4 ms·

My matrix bot that transcribes speech in audio messages

• jacooper 4y ago
  Mozilla has an open source tect-to-speech project, called deep speech. https://github.com/mozilla/DeepSpeech https://github.com/mozilla/DeepSpeech
  • nukemandan 4y ago
    I wanted to use this, but its two tear stale with no release, seems unsupported and likely to fall into being unuseful as other models and tools exponential get better, sadly.
• raybb 4y ago
  Note: this isn't my blog, just copied the title word for word
• IceWreck 4y ago
  Why use google cloud's API when OpenAI released Whisper, which has 2 -5 MB models which are fast af and work so much better than any other models. Theyre open source, run locally and require minimal computation power
  • kixiQu 4y ago
    Perhaps relevant, in response to "Why did you not share this thing": https://twitter.com/maya_Iand/status/1441538613502570498 https://twitter.com/maya_Iand/status/1441538613502570498