What they did? Cleaned data, took a big open speech model (whisper) changed the tokenizer and fine-tuned per data.
89 languages sota speech models. There's plenty of speech data it appears, so the simplest fine-tuning plus tokenization on public data just improves everything substantially. And it's not even with any tricks or all the data... #conll #acl
Users express optimism about BuzzASR's speech recognition advances enabling much better models through collaborative data cleaning and larger-scale training.
No Digg Deeper questions have been answered for this story yet.
89 languages sota speech models. There's plenty of speech data it appears, so the simplest fine-tuning plus tokenization on public data just improves everything substantially. And it's not even with any tricks or all the data... #conll #acl
Call for help: there's a lot more data that needs cleaning, if we do we can train on a scale larger of models and create much much better models. Ones that are likely to be the best in tons of languages for years. Let's help them do that. Reply for suggestions or questions
