YODAS v3: A 1 Million Hour Dataset for the Next Generation of Open Voice AI Research
• 36
AI & ML interests
voice-conversion speech-separation speech-enhancement speech-translation speech-synthesis speech-recognition spoken-language-understanding
Recent Activity
Articles
A collection of models related to the Open Whisper-style Speech Models (OWSM) project from CMU: https://www.wavlab.org/activities/2024/owsm/
CTC-based models from the OWSM project, designed for fast non-autoregressive inference: https://www.wavlab.org/activities/2024/owsm/
Data and models used for EMNLP 2024 Best Paper "Towards Robust Speech Representation Learning for Thousands of Languages"
OpenBEATs: A Fully Open-Source General-Purpose Audio Encoder
The OpusLM collections
A collection of models related to the Open Whisper-style Speech Models (OWSM) project from CMU: https://www.wavlab.org/activities/2024/owsm/
🦉 A suite of Whisper-style models from 250M to 18B parameters. Trained on up to 360K hours of data. 16k sampling rate.
CTC-based models from the OWSM project, designed for fast non-autoregressive inference: https://www.wavlab.org/activities/2024/owsm/
Collection of neural codecs trained in ESPnet for speech tokenization
Data and models used for EMNLP 2024 Best Paper "Towards Robust Speech Representation Learning for Thousands of Languages"