search

rsxdalv / musicgen-prompts

Site for sharing MusicGen + AudioGen Prompts and Creations
https://rsxdalv.github.io/musicgen-prompts/
MIT License
39 stars 5 forks source link

add music sample #1

Closed rsxdalv closed 1 year ago

rsxdalv commented 1 year ago

generated using https://github.com/rsxdalv/tts-generation-webui

JohnTesla commented 11 months ago

Please tall me, M-GEN and A-GEN can use small checkpoint models? (small-med-large music audio) instead of HUGE (many GB!) diffusers in HF/Hub

JohnTesla commented 11 months ago

Can you help me? Why TTS-generation-webui-with RVC Beta inside can work gradio==3.35.2
but Retrieval_based_Voice_Conversion_WebUI .... cannot work gradio>=3.34.0 how to run it with gradio 3.35?

rsxdalv commented 11 months ago

My flight arrives in about 18 hours, I can investigate after that.

On Tue, Dec 5, 2023, 11:09 PM JohnTesla @.***> wrote:

Can you help me? Why TTS-generation-webui-with RVC Beta inside can work gradio==3.35.2 but Retrievalbased.... cannot work gradio>=3.34.0 how to run it with gradio 3.35?

— Reply to this email directly, view it on GitHub https://github.com/rsxdalv/musicgen-prompts/pull/1#issuecomment-1840983329, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTRXIYOFLRD7UJ36TI5NM3YH42I5AVCNFSM6AAAAAAZDORN2CVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQNBQHE4DGMZSHE . You are receiving this because you modified the open/close state.Message ID: @.***>

rsxdalv commented 11 months ago

Can you help me? Why TTS-generation-webui-with RVC Beta inside can work gradio==3.35.2 but Retrieval_based_Voice_Conversion_WebUI .... cannot work gradio>=3.34.0 how to run it with gradio 3.35?

I remember that I always did some work for adapting the other projects (here - RVC) but it might have been quite, maybe as easy as just installing the newer gradio version. In your case, I'd recommend opening an issue on the RVC project and maybe tagging me there since my project uses RVC with 3.35.2 gradio.

rsxdalv commented 11 months ago

Please tall me, M-GEN and A-GEN can use small checkpoint models? (small-med-large music audio) instead of HUGE (many GB!) diffusers in HF/Hub

The models are multiple gigabytes. I cannot tell you the exact reason why it has to be like that, just that audio models are surprisingly large and slow, unlike one might expect given the other AI models.

JohnTesla commented 11 months ago

Can you help me? Why TTS-generation-webui-with RVC Beta inside can work gradio==3.35.2 but Retrieval_based_Voice_Conversion_WebUI .... cannot work gradio>=3.34.0 how to run it with gradio 3.35?

I remember that I always did some work for adapting the other projects (here - RVC) but it might have been quite, maybe as easy as just installing the newer gradio version. In your case, I'd recommend opening an issue on the RVC project and maybe tagging me there since my project uses RVC with 3.35.2 gradio.

Thanks! My idea was: to integrate Full RVC inside the TTS-GEN, Because I wanna train RVC also in TTS-GEN Webui. So RVC full cannot use the same ENV form TTS because of Gradio Ver, so Just need to use a new ENV with gradio 34 for run RVCfull in separated UI

rsxdalv commented 11 months ago

I like your idea. Please create this issue in the tts project

On Sat, Dec 9, 2023, 7:05 AM JohnTesla @.***> wrote:

Can you help me? Why TTS-generation-webui-with RVC Beta inside can work gradio==3.35.2 but Retrieval_based_Voice_Conversion_WebUI .... cannot work gradio>=3.34.0 how to run it with gradio 3.35?

I remember that I always did some work for adapting the other projects (here - RVC) but it might have been quite, maybe as easy as just installing the newer gradio version. In your case, I'd recommend opening an issue on the RVC project and maybe tagging me there since my project uses RVC with 3.35.2 gradio.

Thanks! My idea was: to integrate Full RVC inside the TTS-GEN, Because I wanna train RVC also in TTS-GEN Webui. So RVC full cannot use the same ENV form TTS because of Gradio Ver, so Just need to use a new ENV with gradio 34 for run RVCfull in separated UI

— Reply to this email directly, view it on GitHub https://github.com/rsxdalv/musicgen-prompts/pull/1#issuecomment-1848235160, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTRXI2XQBWY4UUZJ5W6TEDYIPWTLAVCNFSM6AAAAAAZDORN2CVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQNBYGIZTKMJWGA . You are receiving this because you modified the open/close state.Message ID: @.***>

JohnTesla commented 11 months ago

Please tell some more Man. If I want to TTS using trained RVC model (pth+index) how can I do this directly, avoid of generating some speech then transfer to RVC gen

rsxdalv commented 11 months ago

Was planning for that, but you can make a feature request for it in the main repo.

For background - before I went on a holiday I was busy with migratong away from gradio, which I need to build more complex pipelines and better UX; now, I'm still returning.

On Sat, Dec 9, 2023, 10:07 AM JohnTesla @.***> wrote:

Please tell some more Man. If I want to TTS using trained RVC model (pth+index) how can I do this directly, avoid of generating some speech then transfer to RVC gen

— Reply to this email directly, view it on GitHub https://github.com/rsxdalv/musicgen-prompts/pull/1#issuecomment-1848319143, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTRXIYEPRZ52QRSJUVIQXDYIQL3ZAVCNFSM6AAAAAAZDORN2CVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQNBYGMYTSMJUGM . You are receiving this because you modified the open/close state.Message ID: @.***>