Published different RAD-TTS models I've made years ago for Ukrainian to Hugging Face:
https://huggingface.co/Yehor/radtts-uk
https://huggingface.co/Yehor/radtts-uk
huggingface.co
Yehor/radtts-uk · Hugging Face
We’re on a journey to advance and democratize artificial intelligence through open source and open science.
Media is too big
VIEW IN TELEGRAM
Now my space for Ukrainian Text-to-Speech uses GPU, so the generation is fast.
Use here: https://huggingface.co/spaces/Yehor/radtts-uk-demo
#ai #speech #tts
Use here: https://huggingface.co/spaces/Yehor/radtts-uk-demo
#ai #speech #tts
doing something
Now my space for Ukrainian Text-to-Speech uses GPU, so the generation is fast. Use here: https://huggingface.co/spaces/Yehor/radtts-uk-demo #ai #speech #tts
huggingface.co
RAD-TTS++ Ukrainian (BigVGAN v2) - a Hugging Face Space by Yehor
Enter text in Ukrainian and select a voice to synthesize natural-sounding speech using advanced models. The app converts your text into spoken words in three different Ukrainian voices.
Forwarded from Serhiy Stetskovych
Коротше я протестував той елевенлабс ASR. Він дуже дуже гарно розпізнає, якщо брати якісну аудіокнигу то там 100% все правильно, але є одна велика проблема. Він не розпізніє літеру Ґ і замість неї постійно пише Г. Також там тейм стемпи тоже дуже добрі але не такі ідеальні і деколи не попападають. На музиці він працює також добре, але слова які не є літературні може не вгадати і таймінги гірші на музиці.
Now I am working on refactoring of RAD-TTS++ inference code to latest torch here:
https://huggingface.co/spaces/Yehor/radtts-uk-vocos-demo
Code is about 2 years old and I wanted to test how it works on MPS device (Apple M cores).
Inference works on MPS, but not all ops are supported:
By the way, RAD-TTS++ and Vocos are super fast on CPU in MacOS.
#ai #torch #ml
https://huggingface.co/spaces/Yehor/radtts-uk-vocos-demo
Code is about 2 years old and I wanted to test how it works on MPS device (Apple M cores).
Inference works on MPS, but not all ops are supported:
.../torch/nn/functional.py:5561: UserWarning: The operator 'aten::col2im' is not currently supported on the MPS backend and will fall back to run on the CPU. This may have performance implications. (Triggered internally at /Users/runner/work/pytorch/pytorch/pytorch/aten/src/ATen/mps/MPSFallback.mm:14.)
By the way, RAD-TTS++ and Vocos are super fast on CPU in MacOS.
#ai #torch #ml
huggingface.co
RAD-TTS++ Ukrainian (Vocos) - a Hugging Face Space by Yehor
Paste Ukrainian text and select a voice to hear the synthesized speech. Get high-quality audio output with customizable settings for voice, speed, and more.
doing something
No warnings now
Cleaned the codebase, now it's all you need to do Ukrainian Text-to-Speech
doing something
Improved this project: - Replaced ImageMagick by Typst (it has exporting to PNG files) - Replaced Jinja2 by minijinja - Updated Ubuntu image to 24.04 - Updated Python to 3.13 - Added an env var to make Gradio sharing with Docker https://github.com/egorsmkv/pdf…
huggingface.co
Typst-based PDF generation - a Hugging Face Space by Yehor
Enter text to generate a PDF file. Preview the PDF and download it easily.
Forwarded from Feed-Master
Trychroma
Chroma is now 4x faster