SELMA proudly announces that as of this year we are a partner of the SpeechBrain project!
SpeechBrain is an open-source speech AI toolkit created by Dr. Mirco Ravanelli and co-created by Dr. Titouan Parcollet. It’s aim is to making speech technologies more accessible for the community.
SELMA’s contribution to Speechbrain.
- participated to the integration of wav2vec2.0 models (Salima Mdhaffar, Ha Nguyen)
- participated to the implementation of neural Text-to-speech models (Jarod Duret)
- provided recipe for Spoken Language Understanding (Gaëlle Laperrière, MEDIA recipe: https://github.com/speechbrain/speechbrain/tree/develop/recipes/MEDIA)
- provided recipe for Speech Translation (Marcely Zanon Boito, IWSLT22 recipe: https://github.com/speechbrain/speechbrain/tree/develop/recipes/IWSLT22_lowresource)
- And more to come soon!
More about the project: SpeechBrain supports many speech and language processing tasks, including speech recognition (speech-to-text), speaker recognition, speech enhancement, speech separation, speech synthesis (text-to-speech), language modeling, spoken language understating, and many others.
The project has just released SpeechBrain 1.0 in February 2024. More than 140 developers have contributed to the repository.
The code and recipes are available on GitHub. The pre-trained models are uploaded on HuggingFace.
If you want to support the SpeechBrain project and spread the word, give it a « star » on Github!