Видео

I will share when it is finished

Yes it is free. All I ask in return please like and subscribe so I can continue to make more free stuff!😁

Thank you!

Soon? what will it add?

Ah yes, I did noticed this myself. I think there is a way to keep the conversation. In Lm Studio's "local inference server" tab they provide code on how to interact with it. I used the "chat code" but the "ai assistant" code might be the one that remembers the chat history. It should work if the chat code is swapped with the ai assistant code. I will look into this!

hmm you are right. It is doing the same thing to me now too. I will try and figure out what is wrong.

I think i fixed it. Updated the github repository. The program was getting stuck listening for the stop command while the response had already finished. In the basictts_response function after the else statement I added "stopvar = 1" (line 1055) so the listen for stop loop gets killed.

@@mushiekings Thanks I'll give it a shot.

Its working perfect now. Thanks!

Thank you for the feedback! It was hard getting used to hearing my own voice played back and accepting some mistakes along the way. As for your question, I think at the time it required 2 different versions but now it doesn't look like that is the case. Python version 3.10.6 or 3.10.11 looks to be the most compatible and stable one to use with most ai programs. Some sites say to use 3.11 for comfyui to run effectively. Give 3.10.6 a try first and see how it runs. If there's any issues try 3.11

@@mushiekings thanks for the info! I’ll definitely take your advice and see how it works out.

Edge-tts looks like it is online only unfortunately. Faster-whisper looks pretty good. Vosk works but if you don't say the words just right it can have a hard time interpreting. I will definitely check out faster-whisper when I make version 2. Might be the way to go for speech detection. I appreciate the feedback man! :D Now if only I knew how to interact with Applio directly

@@mushiekings im going to try applio, but atm im working on a chatbot with a huggingface model directly with python script instead of relying on lm studio or anything. just python libraries, huggingface, whisper, and edge-tts. i wish edge-tts was local but i'll just set the script up to have it fallback to offline tts if im not online if online then edge-tts. once i get all that worked out, hopefully get a vision model and then i have an idea of how to get it to see what im doing on the screen

Whisper is better than faster-whisper in terms of accuracy. In terms of speed faster-whisper is better

U should check terms of service on their website. As I see u cannot use without their permission

You need RTX GPU to train voices. I believe it has 2 hindi languages if I'm not mistaken.

I'm almost certain you can but you have to be very careful what voices you use. Some voices could definitely get you in trouble. The technology itself is called Retrieval-based-Voice-Conversion. From what I've read there's not much restrictions on how you use it.

@@mushiekings OK, so again it's better and safety to subscribe monthly/yearly to some site (eleven labs, murf,...) and be calm about copyright licenses and commercial use. Thanks a lot.

check the output folder first and see if it worked. Sometimes I get the error also but it still works. "assets\audios" folder

@@mushiekings what should I check for? For the output folder?

@@TonyTonyChopppper tts_output.wav or tts_rvc_output.wav

took a bit to figure that out but basically you want to edit the "run_nvidea_gpu.bat and add --listen "your pc's ip here"--port 8188 and open a spot in the firewall. I'll probably make a video how to do it soon.

hmm Make sure you have a TTS voice and the model selected. Make sure the path file and index file of the custom voice model is correct. Then click on advanced settings and check to make sure the output path for RVC Audio is set. Also check the directory anyways to see if the file is there. I've had a error before(maybe not the same error) but it still made the file. Could be possible you maybe renamed a folder it was installed in so it can't find the path.

FileNotFoundError: [Errno 2] No such file or directory: 'D:\\RVC2\\ApplioV3.0.9\\assets\\audios\\tts_rvc_output.wav'@@mushiekings

i have two computers but This Error only happens on my pc that have a rtx 2060 gpu and not on my old laptop any help? the (tts_rvc_output.wav) isn't being created thats why it shows me an error, theres no file or directory it won't create an audio.

Could try installing the previous version and see if that works@@HyperQuills03

Does it have to do with python?@@mushiekings

Thanks for the feedback. I was thinking the video length might turn people away. Will definitely do that for future videos!

Yes.. sorry about that. I didn't increase my desktop scale. I will remember next time!

Well they just magically appeared after a while. So no need to respond. Now I'm just getting errors every time I try to convert. I'll keep at it.

Not exactly sure where the base voices come from but they should be there when you install it. Also if the base voices aren't quite what you are looking for you can skip the base voice, read the text from another source. Once you have the audio file go into the inference tab, drag it in there and then you render that off a voice model. Thanks for watching😁@insinkproductions

I usually get a error also but it still works. Just check the output folder for the file. @@insinkproductions

@@mushiekings Thanks. I’ll give it a try when I get home tonight. Appreciate your video, and also your response in the comments.

You may have forgot to change the server path directory to the comfy installation base folder. It looks for and installs the models into the ComfyUI\models\checkpoints folder and the loras folder also. Also could try running comfyUI, then go back and try krita again and select "connect to external server" and use the default IP comfyui tells you "127.0.0.1:8188"