Amphion

Amphion is a toolkit for Audio, Music, and Speech Generation.
GitHub:

[Amphion @realamphion]

📚 Explore Now!
🌐 Dataset: https://t.co/NMR7C6LOrZ
📄 Paper: https://t.co/ODeVXYzsCP

[Amphion @realamphion]

🚀Introducing Emilia-Large: 200K+ Hours of Open-Source Speech Data!
We’re excited to release Emilia-Large, the largest TTS pretraining datasets! With 200K+ hours of multilingual speech data, fully open-source. It is ready to use for #TTS and #SpeechLM.

[Amphion @realamphion]

✨ What’s New?
- 2x Scale: Expanded the original Emilia dataset from 101K to 200K+ hours with the new Emilia-YODAS dataset.
- Low-Resource Boost: Enhanced support for languages like German, French, and Japanese.
- Commercial Use: Emilia-YODAS is released under CC-BY

[Amphion @realamphion]

🚀Introducing Emilia-Large: 200K+ Hours of Open-Source Speech Data!
We’re excited to release Emilia-Large, the largest TTS pretraining datasets! With 200K+ hours of multilingual speech data, fully open-source. It is ready to use for #TTS and #SpeechLM.

[𝚐𝔪𝟾𝚡𝚡𝟾 @gm8xx8]

Metis: A Foundation Speech Generation Model with Masked Generative Pre-training

paper: https://t.co/OESKnCYA38
demo: https://t.co/yz8DGDOi1c

[Amphion @realamphion]

@234Sagyboy @discord @discordbots @DiscordBotDevs @_akhaliq @huggingface @reach_vb @Gradio Thanks for the feedback

[Amphion @realamphion]

🚀🚀🚀 MaskGCT!
In addition to the HuggingFace demo: https://t.co/2mCZA9GLzD

you can also join the discord space to play: https://t.co/22IxeaVRq5

Also pre-generated samples: https://t.co/LXiFEiz3Ax

@discord @discordbots @DiscordBotDevs @_akhaliq

[Sylvain Filoni @fffiloni]

Sorry to interrupt but, YES, MaskGCT TTS works for French language !

I have not tested with other latin languages yet, but my guess is that it should work too 🤗

[Sylvain Filoni @fffiloni]

I've added the MaskGCT TTS @gradio API to the Echo Mimic Space, so you can directly clone your voice before generating portrait generation 🤗

Try it —› https://t.co/vSWUt0lbEL

[Amphion @realamphion]

GitHub: https://t.co/zr1hJCWRxa

MaskGCT: https://t.co/ggVXGAQIiC

[Amphion @realamphion]

🔥🔥🔥MaskGCT is hot, making Amphion on the list of GitHub Trending again!

> SoTA TTS model
> Zero-shot cloning
> Emotional TTS
> Multilingual, now supporting English and Chinese
> Fully non-autoregressive and duration controllable

Try in HF and https://t.co/FvmcJ5pm6z

[Amphion @realamphion]

HF demo: https://t.co/2mCZA9GLzD

[Vaibhav (VB) Srivastav @reach_vb]

Fuck yeah! MaskGCT - New open SoTA Text to Speech model! 🔥

> Zero-shot voice cloning
> Emotional TTS
> Trained on 100K hours of data
> Long form synthesis
> Variable speed synthesis
> Bilingual - Chinese & English
> Available on Hugging Face

Fully non-autoregressive architecture:
> Stage 1: Predicts semantic tokens from text, using tokens extracted from a speech self-supervised learning (SSL) model
> Stage 2: Predicts acoustic tokens conditioned on the semantic tokens.

Synthesised: "Would you guys personally like to have a fake fireplace, an electric one, in your house? Or would you rather have a real fireplace? Let me know down below. Okay everybody, that's all for today's video and I hope you guys learned a bunch of furniture vocabulary!"

TTS scene keeps getting lit! 🐐

[Amphion @realamphion]

@reach_vb Thanks to @Reach_Vbarrels A demo from the community: https://t.co/bIRkNKWJud

[Amphion @realamphion]

@Alice2848126245 from the speech prompt

[Amphion @realamphion]

🚀🚀🚀 A Zero-Shot TTS model MaskGCT (Masked Generative Codec Transformer) is open-sourced in Amphion now. Trained with Emilia. Only needs 5 sec speech to clone
Paper: https://t.co/OdoQ3niCeY
HF: https://t.co/2mCZA9GLzD
Discord: https://t.co/FvmcJ5pm6z

Watch the demo by MaskGCT

[Amphion @realamphion]

@Sathees89347227 In a few weeks

[Amphion @realamphion]

@cosmic_spec Internal version is about 500ms

[Amphion @realamphion]

@D3crypTor_X Yes, you can control it

[Amphion @realamphion]

@ElonMuskAOC Your new AI interview 😂

[Amphion @realamphion]

@mohamed17381489 Yes, it supports multi-lingual. We are going to release another checkpoint that supports 6 languages soon.