Voice Cloning For Any Language | Fine-Tuning Tortoise-TTS | Part 2

  Рет қаралды 2,840

Martin Thissen

Martin Thissen

Күн бұрын

Пікірлер: 16
@samasai8860
@samasai8860 2 ай бұрын
Great work Martin! Please make a video on how to finetune Tortoise TTS for our own voice
@tempertephra
@tempertephra Ай бұрын
Der Schritt ab 3:00 "Adjust Interference Code" fehlt in dem google colab link in der Videobeschreibung. Könntest du bitte deinen Tokenizer teilen, da du deine anderen Modelle ja auch geteilst hast?
@tempertephra
@tempertephra Ай бұрын
Edit tokenizer ist im github fork
@Dheekshith-e3l
@Dheekshith-e3l 5 ай бұрын
Great work ! But i have a doubt on where is the voice cloning part there in this video ? Does the speaker name that u gave is fed to the model in the BTS or is that a pre trained voice of tortoise TTS?
@omaribrahim5519
@omaribrahim5519 9 ай бұрын
Bro this is great! Please watch a video on how to collect your own dataset for your language from public data
@ZYJGO
@ZYJGO 8 ай бұрын
Hi Martin, thank you for your great work! it really solved a lot of my confusion, I'd like to know what your final loss is, the model is speaking a completely incomprehensible language after training, and i didn't change any parameter
@nickk1039
@nickk1039 7 ай бұрын
Hi @ZYJGO. Could you find the solution? I went the same way without any changes and after 6000 steps model still speaks the strange mix of languages. Will appreciate if you share the reason.
@ZYJGO
@ZYJGO 7 ай бұрын
@@nickk1039 yes, you just simply need to train more, around 20,000 steps will generate pretty good results, hope it helps you
@nickk1039
@nickk1039 7 ай бұрын
@@ZYJGO thank you so much!
@iqrabatool1814
@iqrabatool1814 3 ай бұрын
Hey. @ZYJGO @nickkk1039 . I trained it for 10000 steps. For the first 5000 steps the output sounded like German language. After that it starts to sound like an incomprehensible language. Isn't it overfitting? Should I continue training or change dataset?
@ondatabletstore6116
@ondatabletstore6116 7 ай бұрын
Hello my friend! in setting if I uncheck the "delete non final-output" option the individual audio files sound bad while the large combined one sounds good. I would like to know if there is a way to make the individual files also sound good?
@ashuu9257
@ashuu9257 9 ай бұрын
can I implement this tortoise tts model locally on m3 pro 14 core gpu without the support of nvidia gpu or anyting like that ?
@TheAfronymous
@TheAfronymous 3 ай бұрын
i have the same question. If you ever have any update, please share it
@StoriesWithAPurpose
@StoriesWithAPurpose 9 ай бұрын
Thank you so much.
@NT3wazLcUqwA
@NT3wazLcUqwA 7 ай бұрын
how abt cantonese language?
How I Built My $10,000 Deep Learning Workstation
22:25
Martin Thissen
Рет қаралды 15 М.
Voice Cloning For Any Language | Fine-Tuning Tortoise-TTS | Part 1
22:53
Правильный подход к детям
00:18
Beatrise
Рет қаралды 11 МЛН
The evil clown plays a prank on the angel
00:39
超人夫妇
Рет қаралды 53 МЛН
Quando eu quero Sushi (sem desperdiçar) 🍣
00:26
Los Wagners
Рет қаралды 15 МЛН
LLaMA & Alpaca: “ChatGPT” On Your Local Computer 🤯 |  Tutorial
13:07
Tortoise-TTS Fully Explained | Part 1 | Architecture Design
19:32
Martin Thissen
Рет қаралды 2,3 М.
Run Vicuna on Your CPU & GPU | Best Free Chatbot According to GPT-4
18:03
Llama 3 on Your Local Computer | Free GPT-4 Alternative
22:30
Martin Thissen
Рет қаралды 32 М.
Правильный подход к детям
00:18
Beatrise
Рет қаралды 11 МЛН