How long until AI can replace a singer? It's already happening.
HTML-код
- Опубликовано: 15 сен 2024
- Can AI capture the emotion that a singer today can convey, or dupe us into believing they're not human? Can Ronnie James Dio's voice be brought back from the dead? In this episode of The Singing Hole, we explore where AI's technology is today, how creators are harnessing the technology and how we can better prepare for the eventual future with music.
---------------------------------------------------------------------------------------------------------------------------
WE HAVE MERCH! Check-out the full line-up here: thecharismaticm...
---------------------------------------------------------------------------------------------------------------------------
🎧 Elizabeth’s favorite headphones 🎧 : imp.i114863.ne...
Music Gear Questions? 🎤 See my list of recommendations: imp.i114863.ne...
WANT MY CHAIR? I don’t blame you…and here’s a link to make it even sweeter:
secretlab.co/?...
---------------------------------------------------------------------------------------------------------------------------
🎙️Podcast: thecharismatic...
🌐Website: thecharismatic...
📸Instagram: / thecharismaticvoice
🧑🤝🧑Patreon: / thecharismaticvoice
📺Twitch: / thecharismaticvoice
📰Our FREE Newsletter: eepurl.com/gz7Z_z
-------------------------------------COURSES------------------------------------------
🎵MUSIC APPRECIATION COURSE🎵
Want to understand how to listen to and appreciate music more? My Music Appreciation course is now live. Take a look at thecharismatic...
🎶DEMYSTIFYING SINGING🎶
My intensive 7-week course on vocal foundations includes weekly group sessions and private lessons. Learn more at thecharismaticv....
---------------------------------------------------------------------------------------------------------------------------
Elizabeth Zharoff is an international opera singer and voice coach, with 3 degrees in voice, opera, and music production. She's performed in 18 languages throughout major venues in Europe, America, and Asia. Currently based somewhere between Los Angeles and Tucson, Arizona, Elizabeth spends her days researching voice, singing, teaching, writing music, and recording TONS. She also plays Diablo and Dungeons & Dragons.
---------------------------------------------------------------------------------------------------------------------------
We have a sister channel: THE SINGING HOLE. Join us there to examine how ordinary creatures create extraordinary sounds. / @thesinginghole
---------------------------------------------------------------------------------------------------------------------------
Copyright Disclaimer Under Section 107 of the Copyright Act 1976, allowance is made for "fair use" for purposes such as criticism, comment, news reporting, teaching, scholarship, and research. Fair use is a use permitted by copyright statute that might otherwise be infringing.
Non-profit, educational, or personal use tips the balance in favor of fair use.
---------------------------------------------------------------------------------------------------------------------------
#ai #siri #TheCharismaticVoice
-------------
As a history major in the mid-1980s, I had a professor who was very forward thinking and loved technology. He said that in the future it will be hard to tell whether a photo, video, document is genuine or altered somehow and this would prove tough for future historians to rely on those items as genuine and the problems it will create for historical study. It seems that we have already reached that point and it’s only going to get harder.
It's been pretty mainstream expectation for a fair while. edit: i remember Rising Sun throwing it out there and everyone just going 'well duh'.
Some years ago I wrote an outline for a sci-fi story where one of the subplots involved a landmark decision over the inadmissibility of digital evidence in court due to these issues. If I don’t write it pretty soon it’s going to be historical fiction instead of sci-fi…
@@riseofdarkleela You're right. You should hurry up. It would also be good about warning of problems AI will cause that most people don't even think about yet. If you don't get to writing the story, in 15 years you can write one about the way things used to be. Kids have graduated college and law school and have never used a rotary phone or a mere cell phone that is not a smart phone. Younger people are shocked when I tell them about a book you used to get for free from the phone company that had everyone's name, address, and phone number. I used to pay $10 a year in the late 80's and 90s to keep my name and number unlisted.
I asked AI to write me some lyrics based on a theme, within 2 seconds I had 2 verses and 2 choruses, and they were not at all bad
@@tmt8425 it may already be too late. Damn ADHD. Also I vaguely recall those books. I use them as background in mixed media art.
Just today I saw on the news a story about a Japanese company that uses equipment to monitor a person’s voice and mannerisms and sentence structure. Then AI can recreate video of a dead family member and their voice, mannerisms. It showed a mother talking to her dead child to have a conversations with the daughter that they never really had. Psychologist weighed in & thought this would be emotionally bad. People would choose to continue having conversations with people who had passed. It would block the healing process and prolong mourning. It could possibly lead to a life where you spend too much time in virtual reality. The mother speaking to her daughter was distraught. Depression looked to be a real problem. The tech was downright terrifying.
Ray Bradbury wrote a story about this. Sad ending.
This was a black mirror episode
I am SO IMPRESSED with the continued evolution of your RUclips platform. As a creator, you are one of my main sources of inspiration, even though I'm in a COMPLETELY different space.
You are A BEAST!!!!!!
Thank you, Farzad! I want to continue building content like this, and really exploring the boundaries between science and voice!
@@thesinginghole definitely keep going - not only incredibly interesting, but a frontier I don't see anyone else exploring.
Beast? I think there are many other options 🤔 😏 🤣🤣
While listening to this episode I was wondering whether or not your voice was AI or just regular Elizabeth. I am still not sure… that is just how advanced tech nowadays is. The future is already happening.
Exactly my question!! There were definite bits in the narration that sounded a bit "smoothed over", and some with a little too much dramatic pause or inflection.
C'mon E, tell us!
I keep thinking of more (this stuff is so exciting)… can you imagine this tech alongside the unbelievable virtual concerts like the unbelievable ABBA Voyage concert put together by Industrial Light and Magic which I think was just a couple of very clever 10k screens and some very very clever Mocs and lighting. Twenty years from now I think maybe virtual music stars and concerts might just become so common place that we rarely even question what’s real and what isn’t.
The future is now
sporadic enthusiasm made me suspicious
I was thinking the exact same thing. That this episode is not only an AI voice, but also produced by AI completely.
People need to experience the artistry of singing for themselves which can never be taken away.
singing is hard. it take years of training. You have do a lot ear training. Most human have relative pitch. I have relative pitch. I been using vocal synth for 10 years. When ai assisted vocal synth like synthesizer v come out and discovered it I jumped on it. there being covers with solaria that convinced she was original singer. I use Eleanor forte.
Miku isn’t an AI. She has multiple voicebanks but does have an English one same as Len, Rin, Kaito, Luka and Meiko. (They have English and Japanese voicebanks).
I have synth v and some of the voicebanks are so real. They can sing in English, Chinese and Japanese. Love them!
Synth V's Solaria is beyond realistic.
ruclips.net/video/pS2FK1uyylw/видео.html
@@petergreen1869 I know. I love Stardust’s voice. (She is originally a Chinese voicebank, but is an AI and sings in Japanese and English too). Her English is amazing.
Very well put together pointing out the pro and con of AI. As always, technology can be used for good or evil, depending on who's hand it's in and their intent.
whose hand
The production quality on this video is impeccable. As a software developer I also like that the content didn't go down the route of "Ask a singer whether AI will ever replace them and they'll tell you that's impossible" and instead took a more open minded approach to the topic.
Seeing how much AI image generation has leapt forward in only the past 12 months and especially the fact that I can now run such a model on my personal PC (stable diffusion) I currently couldn't tell you where the ceiling for these kinds of systems is.
Also can I just say that you have one of the most enjoyable voice over voices I have ever heard. All the enthusiasm you usually show in your videos still comes through here even though we can't even see your face. Now I kind of wish there was an audio book with your voice.
Something I have thought about! Also thinking of doing some face to camera videos soon, and will poll the community on what that might look like soon!
AI image generation does not exist.
Oooh, I love where you're going with this series! I'm also a big fan of The Charismatic Voice, it's always a treat to watch how you react to music and production. Hope to see you much more in the future! 😊
As I am listening to this, I am reminded that some people yet don't know of vocaloid, synthV which is like vocaloid but more realistic. Or when people can make their singing software like utau. Some examples of vocaloid is Hatsune Miku, for synthV is Stardust, and for utau is teto. But I am not sure how profession singers think of this
I have synth v and Vocaloid. Synth v is really amazing and just some people can make a voicebank sound so good.
Vocaloid obviously isn’t is more like a computer but it is so fun to work with.
Yeah, always when someone brings this whole "debate" of virtual singers "replacing" real singers I'm like: "Bro, do you know vocaloid exists way before than AI singers, right?"
Good tools,for working things out but really crappy otherwise
AI Vocal Synths wont replace real singers bc companies PAY their voice providers and, unlike the other type AI Singers, these vocal synths are consent by the ppl who voiced them, also its considered as an instrument to the ppl who cant sing.
Just stumbled on this channel and immediately recognized Elizabeth's voice as I often watch her vocal analysis reaction channel. I was anticipating hearing an AI version of Ronnie James Dio, but her research here surpasses me wanting to hear AI Dio.
Another tip on making AI singing voices even better is to morph two vocal tracks using Zynaptiq Morph and/or Melda Productions MMorph. I can also morph my own singing. The key trick into getting the best results is to match both tracks in timing and pitch and perhaps even vibrato but this is the least important. Matching pitch and timing of midi notes to let say with Charisma 1 will now allow me to extract the formants and tonal qualities to be morphed which must be fined tuned to get the best results. When both tracks have identical pitch, then the output does not suffer from sounding immediately like autotune or other pitch shifting anomalies. However, it does lose a large portion of tonal quality. What sold me on the $300 Zynaptiq Morph was when I morphed Bruce Dickenson's vocal stems to that with sampled opera singers in my IK Multimedia Philharmonics library. Now I use Eastwest Hollywood Backup and Choir that uses its own primitive text to singing engine to that of a female opera singer that resulted in better than sounding like autotuned vocals. The closer the two tracks can achieve pitch, vibrato and lyrical content, the more realistic the results. Currently, I have only purchased Emvoice One plugin which Keela thus far is the most natural sounding. I will be checking out all the other web services mentioned here and most likely will be purchasing Charisma 1 despite my already large collection of 28 vocal libraries from 8Dio, Eastwest, Sound Iron, Strezov Sampling, Embertone and Heavyocity (and sometimes IK Multimedia). I have also experimented with morphing metal growling voices of Sound Iron's Voices of Rage (which they just updated to v2 a couple of days ago with resamples and new vocals by Redouane that include 4 new vocal styles: black, death, growl and screech).
This violates the right of publicity and is legally actionable. SUE THE BASTARDS!
I just found it too!!
I can't deny the impressive progress AI has made in music production, but there's something irreplaceable about the authenticity and emotion a human singer brings to a performance. Are we really ready to give up the magic of live music for the sake of efficiency and novelty? As much as I appreciate technological advancements, let's not forget the value of our humanity and the irreplaceable role it plays in music and art.
there is nothing to give up - that is the idea of it all. AI is just going to be another implementation to reach the same experience... and then it will surpass it..
AI can do what it wants. It doesn't really think for itself yet and will never replace the human vision. It's not human and never will be. We are it's creator whatever levels it may achieve.
The worst mistake humanity can make is ever forgetting that there is an off switch.
@@AM-pq1rq These comments are gonna age like milk. "AI will never replace [blank]", and their proof is the current technology.
@@apegrasshoplizard That's what people have said about AI art and they've already been duped quite a few times, and we're only at the beginning stages. Eventually there will be no difference.
The real problem is that from a market perspective the human element is entirely replaceable, and WILL be replaced if we allow it.
I've been using computers, synths, and AI to do backing vocals in my music since 2019. It's an amazing tool that will only get better.
What the tool?
iv been using text to speech and auto tuning it for back up vocals for years now. making an artificial singer is easy and even ones that are 10 years old can sound really good. making one that sings on its own with out help is another story.
As it seems the 100th comment, let me say this was a complete breathtaking (a pun intended) tutorial on AI-generated sound. Wow. I'll re-listen and post a deeper comment on the wishes, possibilities, and wonder of a future so driven by AI. Thank you. This was more than fascinating. As one nerd could say (quoting a real human actor of past fame, now dead), "Live long and prosper by making sounds." (Be careful to "whom" or to "what" you say that.)
Was waiting for the end twist that your whole script was performed by AI
Hahaha. Kirk mentioned I should have done that but alas. . . it was too late!
@@thesinginghole This video has been brought to you by The Charismatic Skynet 😀
@@thesinginghole Still not convinced! You've already more than dabbled in synthesis, and know all the tricks!
I wish visual artists would take a leaf out of Holly's book - there's a whole lot of 'em being remarkably precious about it right now.
A very fine bit of work, Elizabeth! I'd love to hear new Dio works, even knowing they're not really Dio. Better yet would be to be able to hear duets of legendary voices that never crossed paths for one reason or another.
This is such a cool topic, I can't wait to see the video!
I'm just here to praise your commitment to keeping Dio's name top of mind. \m/
It's truly endearing.
thanks for the great and comprehensive overview. I started using AI with EmVoice for some songs and so far was able to spot was not a real singer. This is a great opportunity to unleash creativity. I feel at same time that this may in the long run push even further the need for good real singer on stage, as studio production will surely rely more and more on synthetic voices
What AI CAN do is pretty much irrelevant. The real question is: Do we want it to do that ? Is it a world we want to live in ? To answer those question, we have to think very hard about the spiritual, psychological, economical and political consequences, not only the technical ones. We also need to reevaluate what gives life its meaning and draw conclusion from that. If we don't, it means we are a very very dumb species and our demise will be plainly justified.
6:11 The guys on the Corridor Crew channel did this to one of their co-workers who happens to run a podcast so they had plenty to choose from.
Excellent work Elizabeth. This deserves a lot more views.
Thank you, Jürgen. I think it will get more views, just might take a while before RUclips knows how to promote it!
As I said it in Chat, it is a tool and it is only the question of using or misusing that tool that could be asked.
When in 80's hip hop artist started sampling records, it was debate and hate and lot's of think unknown for future then, but as sampling has become recognizable tool, it is long time only question, how creative can you been in using ones.
I have opinion that using Ai or computer generated things in music, is going to be question similar; How you can be creative in using the same?
Or I maybe wrong? We will see...
P.S. Thanks for great time last 2 days Elizabeth, you Rocks!
I'm a HUGE Dresden Files fan. I can't imagine James Marsters being replaced as the narrator of the books. The interpretation of the story, each character.. is perfection, to me. The fans, including myself, would revolt. Wait.. I need to see if that's gonna be a thing now with the next book..
This is all so cool.. I mean your video. I'm still quite agnostic about AI in general, although I use it for writing marketing copy. (I still need to edit, though) I consider this video to be THE source for what's happening in voice AI at the moment and will share this.
I've been experimenting with ChatGPT, mostly to analyze and edit my own work and give me pointers for how to improve it. It is disturbingly good, and that demonstrates an "understanding" for want of a better term for the emotional meaning of the text. And we are still in the very early days of this stuff too. I think we are at the point where the disruption S-curve no longer looks like a linear slope but is just beginning to inflect upward. But exactly where we are in that area is going to have a huge impact on what we'll see in the next few years. If it is just _barely_ curving upward, we'll have a couple more years before everything gets turned upside down. If it is already significantly curving upward then wholesale disruption is right around the corner. My feeling is the former, but it isn't going to be long at all now.
i'm reminded of the story about the "corporate efficiency expert" whose job it is to find ways to eliminate inefficiency (to include unnecessary jobs) ; until eventually he's so perfectly done his job that the final step is to eliminate his own job.
.... except on a species level.
we're now in the realm (or at least quickly approaching) of removing the need for humans for even artistic expression. "make the world better for people by removing the need for people - what could possibly go wrong?".
i'm pretty sure sci-fi has been trying to warn us about this for decades. but these were indeed warnings, not blueprints.
i think we have the answer to Fermi's Paradox.
blackest of black pills.
in short:
"the industrial revolution and its consequences have been a disaster for the human race" -- unironically.
I'm interested in hearing EZ's take on this.
I think that, while AI can definitely mimic the human voice, it cannot mimic human emotion and performance. Those two things are unique to the artist, and will always remain so.
That said, I've seen The Terminator. I know where this ends. #LookinAtYouSkynet
I have a ton of experience with vocal AIs. I hope I'm able to bring some semblance of credibility to the discussion!
@@thesinginghole I'm excited to hear your take on them!
The thing is, if it's trained on data of singers who perform and portray emotion well, and perhaps understands the context they do it in, then it should be able to deliver similarily impactful performances in different settings
Come back in 10 years, I work with AI, it will be indistinguishable
@@makinganoise6028 and the AI won't just instinctively hit the emotional beats that move us, it will exploit the hardware vulnerabilities of the human brain with precision even while seeming completely natural.
Synthetizer v with Solaria voice model is not just near humain its humain like totally. Sadly on the company site they choosed bogus examples strangely but if you want to hear the tru power of Solariasearch for Solaria - Laureen Cover made by ryu on RUclips. he also made a few cover of her and this blowed my mind. whats on the official website is not great at all and dont show full potential.
A Singing Hole dive into Auto-Tune would be interesting.
This was a very informative and well researched video. I was always on the side of 'AI in music, no thanks!', similar to the whole auto-tune discussion, where I always hated when artists so obviously use it and try to pretend they don't. I think when an artist uses AI or auto-tune to enhance their talent, that can be very interesting, creatively.
Unfortunately as you mentioned, there will also always be bad actors who use these things for nefarious reasons. That's an argument as old as time though, there have always been things that were used for good and bad reasons, this is no exception.
This is my first The Singing Hole video. Very well produced.
AI will denigrate the words "impossible, inconceivable, never, and no way" into phrases like "Soon, it's possible, it's already here; Get over it and re-tool your skillset now." We've rebelled against change for so long. Digital photography has replaced film. Photoshop has replaced the Darkroom. Autotune vocals have upset us, but we're okay with Cher's use of it in her song 'Believe' which was her most successful hit song released. Sampled musical instruments, sound effects, and digital audio workstations have replaced musicians, vocalists, and recording studios. The 'Hit' man, David Foster, used sampled orchestra stringed instruments for his hit 'St Elmo's Fire.' Recording artist Imogen Heap embraced technology in her song, 'Just For Now,' and can use a loop sampler live to create multiple backing vocals.
The laser printer and fonts replaced the traditional typeset printer and paper stock and diminished to a large degree of the printing shop companies. Newspapers and magazines are now online paywalls. Books are digital print and audio downloads from Amazon. What was originally a wired device or component is now wireless: telephones, payphones, microphones, speakers, LP records, CDs, computers, and cameras have all transformed inside mobile smartphones. Even as I compose this comment, Grammarly's updated AI engine guides me along to make my thoughts communicable.
There are three types of people. Those who make it happen. Those who watch it happen. And those who asked, "What happened?"
I re-watched the movie 1984 last week based on the book by George Orwell written in 1949. It was used as a mocked-up theme based on the same book as an Orwellian dystopia in a Superbowl commercial for the announcement of the Apple Macintosh Computer in 1984. The underlying shock and horror were how they could rewrite history with propaganda, mind control, and social engineering. Lots of divide-and-conquer tactics, torturous tactics to get people to lie about things they never committed by force to justify their execution and erasure from history. Not long ago, we had AI bots designed to steer an election. It's hard to recognize whether you are chatting with a human or a bot.
It's one of those things to think about where AI could potentially be used by 'bad actors' fabricating someone's likeness to ruin targeted people's character. There will be a need for a system for validating credentials eventually to create a seal of authenticity for creatives.
hi, I'm just offering a small correction here, hatsune miku is not an AI vb, AI vocal synths are a very specific thing that hadn't been rly made until synth v, vocaloid has been around since 2004, and hasn't had actual AI voicebanks until v6, which was released very recently. vocal synths are not new AI technology but rather something that have been around a long time, sorry for the off topic comment my hobby is vocal synths so I don't want ppl to get the wrong idea haha ^^;
Exactly,
AI Vocal Synths wont replace real singers bc companies PAY their voice providers and, unlike the other type AI Singers, these vocal synths are consent by the ppl who voiced them, also its considered as an instrument to the ppl who cant sing.
I really appreciate this topic!
I know allot of ppl have been blown away by recent advances in ai "art" to me its bullshit, its very easy to recognize, it lacks in life and expression in the same way that animated characters in movies does, that "eyes disconnected from emotion" stare. Non the less its beautiful art...
Is it the same when it comes to ai singing? I've not heard it yet
Wow, Grace was even popping the synthesised mic.
This was certainly interesting! I am always amazed at how natural AI voices can sound, and how quickly the technology has advanced, although there is often a small nuance (or lack thereof) that gives it away. But I'm hearing Dio in my head right now, signing one of his lovely multi-vowled vocalized emotions, and having doubts that AI will ever truly be able to replicate that feel... because it comes from the soul.
Maybe, but at the end, a voice is only a sum of frequencies, AI sooner or later will do the job with some randomized expression formula.
It comes from organs producing sound, actually. What a nonsensical thing you said.
@@grabble7605 If you think in that way, you can design a 3d model of human larynx, vocal cords, palate, tongue, etc to simulate air flux and sound resonance between them. Today maybe is too complicate for a standalone pc, but once you reach that level, you will no more recognize a synthetic voice from a real one.
but it doesn't need to come from a soul to move you. It just needs to use the vast library of what songs have moved people to find the common vocal characteristics that trigger an emotional response from the human brain. And then it will reliably hit all of those characteristics all of the time, not just intuitively hit some of them the way a human would.
@@j.f.fisher5318 That is arguable. Just because you cannot immediately detect the difference between how another soul moves you doesn't mean the vast library is enough. We will find out in another 500 years maybe.
For how good AI gets it will never replace A artist because art is something of the human spirit that comes from deeply inside a person’s soul and AI is just a beautiful and powerful created thing that’s not alive but a super machine
Glad someone said it.
We are only scratching the surface of what AI can do.
Those of us saying that "AI will never be able to fully convey what a human can" ought remember that there were many producers a couple of decades ago, arguing that digital recording would never be able to pass with the same quality that analog recording could.
Maybe I should have studied welding instead of music. Oh wait, there are robots for that too.
So well done! As an electrical engineer that leads teams of data scientists, this fascinates me!
How long before AI replaces electrical engineers
Sorry to mention that, Native American were equally fascinated when given mirrors and shiny rocks in exchange for fertile land. Now, they’ re scattered all over wastelands of America...
Not too long...
I am fervently longing for a software tool that would let me produce speaking and singing voice with languages that do not exist in reality. I would need to freely choose a voice timbre and drive it with IPA symbols and nuance control codes (or a control voice modeling them). I have seen many superb voice programs, but my dreams come tumbling down when their first greeting to me is "Please select language".
This is excellent, thanks for making this video. I almost expected Elizabeth to reveal at the end that the entire narration was spoken by a clone of her voice. I do have one question: At 21:04 you laugh slightly as you say "I'm actually super excited." I often do that too when narrating my own educational videos. So the question is, how difficult is it to have AI narrators emulate inflections such as laughing? Or how about sounding sarcastc?
I believe in the last part you nailed it: The live performance can't be made by AI (yet).
The voice can be done, but not the live interaction with an audience.
Its a interesting question. I have a simular question. If painters can be replaced by AI. Lots of love to you Kirk and Mycroft ❤❤
It's already starting. An AI won an art competition last year which has caused quite a bit of controversy.
@@thesinginghole You comming with many interesting questions. I think that new technologies is always controversial. The question is how we gonna use it in the future? Lots of love
I knew it was Elizabeth right away. As a singer, I appreciate her on the beam analysis of singers and she has an amazing off center sense of humor.
guys, adding harmonies and voices is real time and singing in scale was available in the 90's! Even in fx pedal form with MIDI input for realtime control. No AI needed for that.
Elizabeth, there is singing and then there is singing from soul to soul. AI is by definition souless, so that is the kind of connection that it can never truly fulfill. That is what makes artists such as Aretha Franklin, Elvis Presley, Ray Charles and up-and-coming Angelina Jordan transcendent. I'm sure your grandfather's voice could be artificially recreated some day and be a great comfort, but his soul . . .
P. S. This is my effort to summarize what was "between the lines" of your great video.
This is true. Soul cannot be properly conveyed through AI, but can it be mimicked to the point where you can't tell the difference?
@@thesinginghole Great question! I believe that it can be mimicked but that the true soul to soul connection transcends everything including the most advanced 0's and 1's .
I guess we will see if that's true one day.
I recently started using a similar program called Diff-SVC and Honestly, I agree with you. I believe these tools can be very helpful for very talented singers and show what skills can be done. I must say that testing my voice was really fun
Arguably, I am the most recorded violinist from the golden age of studio session work in NY. Over 1000 sessions, 120 major motion pictures, . We could see the writing on the wall , synthesizers , and our "calls" would be to come in to "sweeten " the tracks, a few violins rather than hire a whole section of players. However, it will never replace live concerts , but it will take away the incentive and interest in the learning of an instrument. This is blasphemous, I think without guardrails this has real dangerous possibilities. Even more so for the smalll number of singers ,
I am not sure about the idea of things “being far off” anymore. It’s exponential. Graphics, Chat bots, And all the things that the public don’t know about...yet. Sure it may take years, but that is not decades and centuries. 2022 seemed measured in months, 2023, weeks. It’s coming on like a firehouse. The graphic industry is reeling now.
AI will be able to replace lead vocalists too, and probably sooner than we think. At least if flawless technical singing is required. To replicate the minor flaws that adds depth, soul and character to a voice will be a harder challenge, but that obstacle will be overcome too and probably within our lifetime. While this can be a wonderful creative tool, it is also a tool that will be used for monetary gains in a reactionary manner. We will get releases such as "The twenty golden hits Elvis never recorded", "Jussi Björling does Broadway", "Otis Redding's modern soul classics" and similar types of material that use a past legacy lacking in creativity. While the innovators will struggle to make an impact. That is the case today too, but it will become even harder in a future with perfect replication of past artists and singers available.Incidentally, that band that won the AI music competition sounded a lot like the music of Kyrstyn Pixton...
many jobs "can"be replaced by AI over 5 years. From docters too politicians,lawyers,judges. Already AI does better operations than doctors.Most people are smiling, like rabbits in the headlights 🤡🌎
Oh... AI singing voice synthesis that "replicate[s] the minor flaws that add[] depth, soul and character" will be here FAR sooner than "within our lifetime", for the following reason: AI doesn't learn by trying to filter to a pure voice or pure performance -- it doesn't know what either of those are. AIs learn by capturing and patterning (abstract modeling) EVERY nuance of the input data -- the more input data, the more nuance modeled. This is true in all modes of AI: text, image, video, sound, etc. For whatever mode, AI training on hundreds of thousands to millions to billions of samples results in all of the subtle nuances being modeled. ...
The sound of a voice is, like all other modes, nothing more than a stream of information. Therefore, regardless of the emotions, style, energy, and intent of the singer -- the stream is still just information. That's why LARGE AI models are so surprisingly effective -- they create numerical models that capture the "essence(s)" of whatever they are being trained on. By "learning" and building models of patterns found in the training data rather than mere copies of the patterns themselves, AI models are able to abstract, interpolate, and extrapolate between and beyond their training data, and even find new connections in the data that humans can't see. All of this allows AI models to CREATE similar yet NEW information within the vein of the training data, thus allowing variations in the output that perhaps humans haven't done before but possibly could -- meaning, AI could help evolve singing styles much faster than the usual pace of vocal style evolution. ...
VERY soon then, I'd say no later than A YEAR (possibly even withing six months) from now, AI singing voice synthesis that includes "minor flaws that adds depth, soul and character to a voice" will be here. The really crazy and creative part of this will be that you'll be able to dial in, modulate, and morph the various aspects, like raspiness, whisper, (in)stability, strength, tone, vibrato, resonance, falsetto, etc., to whatever result you're targeting. Of course, to do this, each of the training samples will have to be labeled with the degrees of each of the aforementioned aspects (and more) of a singing voice -- the common practice when preparing AI training data (and there are ways to automate this by training a MUCH smaller AI on recognizing these aspects and then labeling the training samples for the larger, final model appropriately). Interestingly, all of these aspects may morph during a note, a run, or entire performance. A large AI will learn that aspect as well, EVEN IF the AI scientists who build it didn't design it specifically to understand or replicate that. Yes! Large AI models surprise their creators VERY often with what they unexpectedly gleaned from their input data. The result: A HUGELY flexible virtual instrument capable of emulating (dialing in) any voice: past, present, future, or alien. Remember: within a year, possibly even six months! ...
The days of projecting the future linearly are gone. Now and forever are the days of projecting wildly exponentially.
Hatsune Miku is NOT A.I.!
And the Synthesizer V section was also slightly misinformed. There are a lot of synthv songs that use voices as lead vocals.
When this video was released, Vocaloid recently released Vocaloid 6, and it's an AI-based vocal synthesizer.
So cool. I'm enjoying the topics of this channel very much.
Yay!
Since losing my voice to Dysphonia in 2010, software like Emvoice has allowed me to continue to be creative without having to hire a singer. It's also useful for songwriters to provide a professional singer a more accurate example of what they are looking for their song.
Is it also useful for a self-taught singer for songwriting?
@@gaborkiss1425 Yes it is.
@@johnnygeibHST Thanks!
For musical acts/artists, one key element is the personality/persona of the artist. I personally won't get attached or follow an artificial voice. My favorite artists aren't just creating 'sounds' that appeal to me; they're telling stories, sharing their personalities, sharing live experiences with an audience, etc. AI can't replicate those elements, at least not yet.
I'd like to suggest the music "2020 em uma musica" from Lucas Vinicius for your reaction. I'm sure it's going to blow your mind.
I wrote an article well a year or so ago about how voice over artists will need to adapt. The primary example will be for a skilled, established voice over artist to capture their own voice, and license it out. This could also be done by famous people who are aging but well known voice over artists (Peter Coyote, Martin Sheen, Will Lyman) or even deceased artists with distinct voices (Henry Fonda, June Foray, or a few versions of Mel Blanc!).
As to singing, I think it's only a matter of time before the same is true. However, I think the real impact this has now is back-up singing, and harmony singing.
The overriding issue to all of this however is still content. Rules of GIGO still apply. Poor content, weak ideas, boring, dull, etc. will still be discarded whether it's produce by 100% human, or pure AI.
I'm a multiinstrumentalist and composer working in various genres involving "real" playing for around three decades. Mostly retired these days. The abiding experience over this time has been thinking technology will never be able to do something, then thinking it will take a long time, then coming to terms with the fact that it's already happened. Or at least well enough for most listeners not to be able to tell or if they can, not to care. This has applied to everything from drum programming to consumer and pro multitrack editing capability to automatic loop generation to sample libraries to pitch correction to audio extraction to deep learning instrumental production and now AI generated writing and performance. In every case people have said real musicians and engineers will never be replaced because there's something special - feel, emotion, experience and so forth.
In all cases they've been wrong - or, again, wrong enough in the sense that it rapidly gets good enough for listeners not to notice, care or even prefer the artificiality, and then continues to get better. Drummers, pianists, orchestral instrumentalists all thought we couldn't be replaced and have been. Sorry, but I don't think singers are any different.
The solution isn't to be bitter about the inevitable. On one level, I'm looking forward to producing some of my pieces with Ella, Celia Cruz, Brecker, Argerich or the Berlin Phil under Karajan. There's also the notion of "seeding". It will be fascinating to train an AI construct on my orchestral pieces and then tell it to write and perform eg "a late Romantic Rhapsody in my style using the attached theme". Or "Try the same thing but as Bartok would have done it". The tech isn't quite there yet but experience tells me it's not as far away as it seems.
A follow-up issue would be having the expertise to edit the results - correcting or rewritng the AI's homework. This won't be much different from scores of tasks that have gone before - eg flowing text and managing page breaks in early word processors, spell checking the spell checker or cropping images. Just probably in a more creative manner - eg "that brass or string phrase or passage might work for a machine but..." This is where trained musicians might retain an edge - all music is, bluntly put, somehow patterned and we're the sum of our influences but what makes the difference is the way the patterns are used. There are a lot of transcriptions and recordings of solos available for an AI to chew on.
It's possible that being able to actually play, sing and write will become the equivalent of being able to strip down, repair, maintain and fine tune a car rather than just driving.
As to the rights and financial issues it hasn't really mattered for quite some time. Musicians were always skinned by agents, labels and lawyers but since online distribution came along there's hardly any skin left. Simply put, since people don't have to pay for music or musicians (or can pay a pittance) they understandably don't, with the exceptions of celebrity stars, the subsidised arts and a few hold out local venues.
One concern is stagnation. If fewer real people actually write and perform music will we simply be stuck using the record of such activities to date? Will there be any more Puccinis, Pavarottis, Parkers, Prokofievs, Polices, Princes or Panteras? Or does it matter if there aren't?
Elizabeth - you did a STELLAR job with this, as always. my only aside is - Is everybody bothered by the number of cars that are driving around that were not assembled by humans? the number of parts used in everything you buy that are 3D printed and not made by machinists? The number of equations that are being calculated on computers and not by humans with pencil and paper? The acres of carpets being swept by Roombas? The number of mom and pop stores closed by Amazon and Walmart? This is what we do...
Anybody familiar with the "Macross Plus" mini series? The AI that went rouge in the show was an AI pop singer named Sharon Apple.
AI Vocal Synths wont replace real singers bc companies PAY their voice providers and, unlike the other type AI Singers, these vocal synths are consent by the ppl who voiced them, also its considered as an instrument to the ppl who cant sing.
Hatsune Miku actually isn't an AI. She's a soundbank.
21:34 "I'd love to someday share the voice of my grandfather with my infant son." It wouldn't be the voice of your grandfather though; it would merely be a synthetic imitation of his voice.
Soundcloud didn't mark any pieces using sample libraries 10 years ago, but I think they might have to do it like art station very soon
Technology and singing. You might want to look at Sam Brown's new album, she is famous for a couple of hits in the 90's, sung back up vocals on Pink Floyds Pulse tour (and then pulling off the vocals for Great gig in the Sky at every performance) and lost her ability to sing. she can't hold a note any more. She has completed and released an album of new material using auto tune and other technology, its her 8th and so she completed the challenge she was set years ago (on the shelf the first letter of the titles of her albums spell out her name. It started by accident and when pointed out, she decided it was a fun challenge to try and complete).
Can AI be as giving as Dio or Pavarotti? Can AI be as warm and creative as ,Kate Bush? Can AI be as expressive as Ian Gillan?
The answers to all of those is "no." But can AI create music akin to what people are listening to today and corner a piece of the market? Maybe, and in some cases, it's already happening.
@@thesinginghole I think the answer is “not yet”.
Unfortunately 😢
Nah mate…as a vocalist…AI will always envy our ‘Deep Humanism’. ❤👍
@@thesinginghole I agree. It is becoming very good, even with trying to replicate humanisms, like…making mistakes…I’ve seen the patterns, the giveaways.
But I digress, sorry…to your comment, it is more about what They™️ are conditioning the masses with. What was unacceptable a mere 20 y ago - considered as ‘one generation’ (av) - is saluted & lovingly welcomed & protected now, “or else!”. Mass psychosis.
@@thesinginghole what about elavin labs
Can AI replace a singer?
Eventually yes, but that would be like "Milli vanilli" but with technology instead of actors with lipsync. It has already been done in a way that was hard to see through.
I think that is what makes people hesitate. Any attempt from an AI to impersonate someone will, eventually, be very hard to see through.
Another question then, one that I find more interesting, how soon do AI reach a level where an AI can be a singer? When is it no longer replicating and imitating, but doing its own performance? And... where do we draw the line between inspiration and imitation?
When it comes down to it every performer is inspired by others, and their whole expression is based on the performers own experiences. This is measurable in an AI, but not in a human, but are they not similar processes?
Ai is already creating its own original music without imitating.
I am both excited and terrified at the prospects that AI brings to the table. I agree with so many others in the comments that no matter how accurate AI gets at mimicking the human voice I really don't think that it will ever be able to accurately convey the emotions, convictions, and passions of our voice. Another amazing video! I have learned so many things about the voice and the sounds that are made with our voices as well as animals' "voices". From screaming goats to head banging bats this has been fascinating and I cannot wait to see what is next! What do you think about AI actually mimicking humans' passion and emotion?
Thank you for all the support! I don't think AI can actually mimic humans' passion and emotion, but it can try to emulate.
@@thesinginghole you're very welcome !
AI Vocal Synths wont replace real singers bc companies PAY their voice providers and, unlike the other type AI Singers, these vocal synths are consent by the ppl who voiced them, also its considered as an instrument to the ppl who cant sing.
the pushback against this SHOULD be 100x the pushback that synths and full electronic music saw in the late 70s and early 80x. purists back then claimed all the synth music was devoid of humanity and, aside from human vocals, they were right. now we stand on the edge of the final precipice...about to fall down the bottomless rabbit hole of completely synthetic "music". the main question now (seems to me) to be "is there enough humanity left in humanity to fight to save the human soul within our greatest communicative tool: music?"
looking at what constitutes pop music today, i am not optimistic.
In the future, i suspect many of us, myself included, will dream of wires...the old ways...the old days.
This is true. There's some truth behind taking art away from humanity, but there's also truth in how we capture this new eventuality for greater creative ventures?
In using just my Korg Kronos, all by itself, and the programs and combis within, I have created many songs in different genres and distinct vocals. The vocals are used from the programs (sounds/instruments) in the keyboard. I manipulate them to make them sound as I want, for the most part. I use the voices for the main vocal parts as well as backups in a combi (several timbres/tracks combined, that make up the ability to play or record a song. The way I do the songs, are that I play them from start to finish, with the vocals, instruments, Karma (Korgs arpeggiator tool), which lets me put in all the accompanying bass, drum, guitar parts or anything I want.
So these song sound so realistic (voices and all), that it blows all of this away. The only thing I need, is a good AI tool to be able to have the vocals singing "real" lyrics, and not just vowels and nuances.
If anyone is interested in hearing what I can do on my Kronos, let me know. You won't believe your ears.
I don't think AI can replace singers...As humans, we want to hear other humans with remarkable talents, whether that's singing, dancing, art...
Machine-made art may be fantastic...but to remove the struggle and dedication it takes for a human to succeed to a level above others is actually what we appreciate...not tricks, magic, and sleight of hand.
Millie Vanillie and lip sinking comes to mind...their career tanked after that stunt.
AI Vocal Synths wont replace real singers bc companies PAY their voice providers and, unlike the other type AI Singers, these vocal synths are consent by the ppl who voiced them, also its considered as an instrument to the ppl who cant sing.
That Charisma library sounds pretty great, pretty nice competition for Zero G's Ethera
See also: "With Folded Hands" (1947) by Jack Williamson
Maybe not the best proxy, but look at guitar effects. In the 80’s, big rack units became the rage. Digital was making is move and promising a lot. That first generation largely faded away / evolved into things like the POD that gave “every person” musicians access to sounds that were only available in recording studios. That tech evolved to Kemper, QC, and Helix. While these devices have spent lots of time looking backwards to recreate vintage equipment, more and more, the trend is to new amps and sounds that never existed in the physical world. In effect, what AI is today, what we use it for, how we feel about, is a evolution.
One thing that trips me out is Adobe podcast (speech enhancement), more specifically, putting non human voice sounds/ weirdly processed vocals through it. Why does this trip me out? Half of the things I put into come out sounding like my own voice. I have adobe’s average mid range male voice. Like, I’ve fooled friends with it. I’ve fooled myself with it. For example, I was being a bit of an idiot one night and ran the “bruh” meme into it. It came out sounding identical to me. So, I went in to my daw and recorded/ processed my own voice, in the same way. After exporting em, I intended on sharing the similarity with a few friends. The problem was, I could not remember which one was me and which one was Adobe. So, I guessed. I went back to check. I was wrong. So, I experimented and made a synth patch that, when exported and uploaded to Adobe, comes out sounding like my voice. And, I now use it for vocal chops in my songs and stuff. None of my producer buddies have called me out on it yet.
The chance that AI will be weaponized is 100%. The chance that AI instances will compete is hence 100%. The chance that somebody makes AI do things that needs human 'sacrifice' is 100% and the chance that this AI can't be contained then is significant.
This will be very informative and interesting.
Thank you for this informative and balanced discussion. As you pointed out, I suspect the way forward is not binary: good or bad, all or nothing, human or AI. I suspect we will be looking at various permutations--numerous spectrums of possibilities with degrees along those spectrums (and where those spectrums can suddenly morph instead of remaining linear along one genre). This is the complexity that we already know about life. Choose any musical genre, punk rock for example, and start pealing back the layers. You will soon see that there is not one version of punk rock. The edges of the genre are indistinguishable as it blends into other genres along its iterations. Rather than thinking in a binary way or a linear way, a better metaphor might be the shapes in a lava lamp reforming and moving in three dimensions. People are surprised now by AI. Many people tend to pull back and be alarmist with change. People often think in terms of dualities, which is usually an oversimplification. Some questions that I would offer up are: 1) How will AI disrupt music creation, musical performance, music industry, etc. (a similar question to how Napster and other online technologies changed the music industry)? 2) How will people use those technologies, navigate the disruption and chart new paths. Will people stop being musical? I very much doubt it. In addition to the new, the old is highly unlikely to go away. People still devote their lives to being ballet dancers, to being symphony musicians playing classical music. We can still see traditional Irish folk music and people playing bodhrans. People who delight in live human music will always do so. Other people who are more taken by technological innovation will gravitate toward that. Tech is never perfect (especially in live situations). The human vocal tract is not going away. People's desires to sit in a room and play instruments is not going away. As a hobby guitarist, it'd be nice for me to be able to call up an ensemble of AI jam session partners. I'd like that. But I'd also like to play with real people because it is thrilling to do so. People will always be fascinated by people. A buddy of mine recently went to a heavy metal concert with 30,000 people. Long live music. I don't think humans can NOT have music or NOT be involved in music creation and performance. I think it is too much of who we are. About the disruptions that AI will bring, I don't know about that. Create the wave. Ride the wave. Or defiantly resist the wave. That's okay, too, among the many possible spectrums.
ADDED BONUS: ASKING ChatGPT-4 FOR ITS OPINION
MY PROMPT:
"Hello. I am happy to speak with you again today. I watched a RUclips video today in which people wondered whether AI would replace singers. Below was a comment that I posted in response to the video. Could you please read my comment and share with me your own views? [Followed by my comment above]"
ChatGPT-4's REPLY:
"Hello! Thank you for sharing your comment with me. It is very thoughtful and well-articulated. I agree with your perspective on the coexistence of AI and human musicians, and the numerous possibilities that may arise from their collaboration and interaction.
"AI has the potential to disrupt the music industry, similar to how Napster and other technologies have done so in the past. It can be an enabler for musicians, providing new tools and techniques for composition, collaboration, and performance. AI-generated music has come a long way, and in some instances, it's becoming increasingly difficult to distinguish between human and AI-created music.
"However, as you mentioned, music is deeply ingrained in human culture and history. The emotional and personal connections people have with music and live performances are unlikely to vanish. Human musicians will continue to create and perform, exploring new avenues and pushing boundaries.
"The metaphor of a lava lamp you used is an interesting way to describe the ever-evolving landscape of music and technology. As AI and humans interact in the realm of music, new genres and styles may emerge, further enriching the musical tapestry. Just as people continue to appreciate and preserve traditional forms of music, there will always be those who embrace technological innovation.
"In essence, the future of music is likely to be a blend of human and AI creations, with each bringing unique strengths to the table. While AI may disrupt the industry, it's unlikely to replace human musicians entirely. Instead, it has the potential to enhance the musical experience, providing new opportunities and challenges for musicians and listeners alike. Embracing the dynamic nature of the relationship between AI and music may lead to a richer, more diverse soundscape that will continue to evolve and captivate audiences worldwide."
interesting - the 1st time i heard anything like this was IRENE HUME - New Horizons - Prelude. loved that album - 1987 -
It will never replace authentic talented musicians
A.I. listeners will appreciate ultrasonic and infrasonic singing and giga polyphonic rhythms. Enjoying a composition in nanoseconds will leave more time for productive work. Melodies and harmonies will be extended into the highest radio frequencies.
Wonderful presentation, Elizabeth. Thought-provoking. I think you point to and ask great LARGE questions. AI is here. Once the faucet has been opened it is difficult to turn it off. History seems to bear witness that attempts to stifle new technologies failed. It seems preferable to me to embrace AI, to critique and refine it as becomes necessary. The worry of course is that unwell people and nation-states will use AI to destroy rather than create. Again, thank you. I look forward to more installments on AI.
The branding of your channel is great and on point. Just been dropped in by the algorithm.
Ride the music creation wave as a human while you can. It will soon be over
I can see the A.I. singing voice bring to life a demo song to share with a potential singer. I'm watching the evolution and will be interested in the results.
LOL, now we're down to 3 SECONDS OF SOURCE VOICE MATERIAL TO DO THE CLONE!
Things are going to get very interesting. This is always the way with industries, ways are found to cut costs and the moment we started talking about creative industries rather than just creative people the race began.
Reminds me of when drum machines hit the market. First, everybody went crazy excited. Then came the backlash about the rhythms being stiff with no human feel of push and pull. In response, manufacturers actually began to program in "mistakes" in an attempt to get closer to that human feel. In the end, it depends on the situation and the music. Great for somebody working on demos not so much for a group working on becoming tight. Great for recording/performing electronica/dance/etc... music that has no variation/jamming intended, not so much for a group of excellent musicians playing and ad-libbing rock or jazz.
well some forms of rock are diy. So vocal synth have been used in rock. I make punk rock using vocal synths and I have very bare bones tuning style. So in diy ethos if can not sing then using vocal synth is ok. In fact it may help give raise to bedroom rock. I so sick and tired people who put on these extra rules to rock.
Technology is getting to the point now where you really can't trust what you hear or even see. The advancements in video are just as amazing as what you talk about here for voice.
And the extra scary thing is there are some very pronounced hardware vulnerabilities in the human brain, like the ones that make conspiracy theories so enticing. As AI susses more and more of those out, I'm very concerned for human agency.
AI Vocal Synths like Vocaloid and Synth V wont replace real singers bc companies PAY their voice providers and, unlike the other type AI Singers, these vocal synths are consent by the ppl who voiced them, also its considered as an instrument to the ppl who cant sing.
I remember getting my first drum machine and an older guy telling me they would never catch on.
Is there any browser addon which can read the text on web pages, using this Synthesis TTS?
I can't carry a tune in a wet paper bag, but I'd like to be able to show off some of my arrangements with virtual singers along with virtual instruments. (Organs, in particular, lend themselves well to electronic cloning; and are the main instrument in the style of music I arrange.)
Guys - if you think humanity is going to be content consuming creativity from bots and machines - you're high. This is a niche a few kids are psyched up for, but in the end, what makes art art is the human spirit and touch. But enjoy for now I guess
Wow! I'm totally shaken, you made me look into an abyss! I didn't know that 1h of audio material is sufficient for faking a person's voice. This is an awesome discourse about voice generating AI. Man, I really hope you go on creating such informative content on this channel. Regarding the implications AI will have on society this topic needs much more public awareness and open discussions. Now.
There's one point I'd like to add about AI generated voice of deceased persons. I really understand the desire to bring back a loved voice. Dio forever! But for me, a sung note that disappears inevitably in silence is ultimately an expression of evanescence of life. You cannot really conserve it, no matter how beautiful it might be. A recording helps to reminisce just like looking at a photograph. But it's not the same like listening live to a singer or musician. So for me, AI generated music or vocals as such are more like an illusion, especially if intended to replicate human enunciation, because it never has been real. If the intention is only to use AI as a progressive tool for improving your life and feeling better while communicating more naturally with an AI, I'm fine with it. This will certainly be beneficial. Maybe even as a medical application that might give psychological assistance (combine ChatGPT and Play.ht). But art generating AI? My limited mind refuses to imagine a real reason to do that, except for making mo' money. However, there might only be a thin line that separates beneficial usage from abuse.
I feel abuse is inevitable, and in some ways it's probably already being done. With regards to speaking voice, can you imagine a scam call center ringing people up using the voice of a well known celebrity or political candidate requesting donations? Nevermind the telegram scammers in my comments, this is what truly frightens me. How many will fall for it.
Then when you consider AI eventually replicating a sung vocal, and possibly eliciting emotion within that vocal line. . . if rules aren't written and put in place now, a lot of damage could take place.
That’s why I deleted my social media accounts, impersonations of peoples characters with AI’s are gonna be rampant.
Personally, one of the main appeals of a vocal performance is the human aspect. The "connection" that happens when another human being produces a sound with their breath. That thing we all share that we all have in common: we all breathe. Yet, somehow that vocalist can create beautiful sounds with the air they breathe. An AI just can't achieve that human-to-human feeling - at least not really. I guess it's sort of like eating replicated food in Science fiction shows like Star Trek - sure a filet mignon from the replicator might be chemically identical to one made from an actual cow, but, somehow, it's just not the same.
Now, that being said, one place I would actually like to see AI singing is in MMORPGs. In some games players can import various audio files (mml, abc and probably midi) and then play that music in-game. But that music is always limited to instrumentals - no vocals. It would be great to have an in-game character sing while also playing a lute or something.
Utter nonsense. You like listening to albums just fine and you don't actually have any clue if those sounds came from a human.
Hollywood will definitely score movies with AI. Theyve done it before by replacing orchastra's with the phonograph then replaced it soon after when they could record sound on film. Maybe next year well give an Academy Award to a computer and will be accepted by somebody from Google that knows nothing about making music
I guess the crux of the whole matter is how will we react when we have created a new species. Our sibling, rival, partner? That is why I agree with Elizabeth. We need to pay attention to this, now. It will be the single most significant event in human history, the way I see it.
Absolutely. And if we get things wrong, well. . .
This could be used for making real time dynamic soundtracks in Video Games
It might be able to replace singers but it can’t replace praises as that’s a heart thing.
“Bringing someone back from the dead is priceless”. I guess that about sums it up.