search

FL33TW00D / whisper-turbo

Cross-Platform, GPU Accelerated Whisper 🏎️
https://whisper-turbo.com
Apache License 2.0
1.67k stars 68 forks source link

[Question] Support Large Model #31

Closed Mineru98 closed 10 months ago

Mineru98 commented 11 months ago
  1. I don't think there is a binary file yet because the source code seems to support the Large model. Is it currently being built?
  2. As far as I know in openai, there are whisper large v2, are you planning to make large v2 as well? Looking at this link, it's only possible up to v1.
FL33TW00D commented 11 months ago
  1. I don't think there is a binary file yet because the source code seems to support the Large model. Is it currently being built?
  2. ~As far as I know in openai, there are whisper large v2, are you planning to make large v2 as well?~ Looking at this link, it's only possible up to v1.
  1. Large model is going to be feature gated for enterprise only. Once we fix some bugs and ship quantization Large-V2 will be available.
  2. I've implemented the runtime myself, no whisper.cpp so any restrictions they have are N/A.
Mineru98 commented 11 months ago
  1. I don't think there is a binary file yet because the source code seems to support the Large model. Is it currently being built?
  2. ~As far as I know in openai, there are whisper large v2, are you planning to make large v2 as well?~ Looking at this link, it's only possible up to v1.
  1. Large model is going to be feature gated for enterprise only. Once we fix some bugs and ship quantization Large-V2 will be available.
  2. I've implemented the runtime myself, no whisper.cpp so any restrictions they have are N/A.
  1. Enterprise only is welcome. 😄
  2. That's good too 😄
  3. If the large model is improved, will you write a contact somewhere in the repository?
FL33TW00D commented 11 months ago
  1. I don't think there is a binary file yet because the source code seems to support the Large model. Is it currently being built?
  2. ~As far as I know in openai, there are whisper large v2, are you planning to make large v2 as well?~ Looking at this link, it's only possible up to v1.
  1. Large model is going to be feature gated for enterprise only. Once we fix some bugs and ship quantization Large-V2 will be available.
  2. I've implemented the runtime myself, no whisper.cpp so any restrictions they have are N/A.
  1. Enterprise only is welcome. 😄
  2. That's good too 😄
  3. If the large model is improved, will you write a contact somewhere in the repository?

I’ll keep this issue open until I am happy with the state of large-v2, and will let you know when it ships.

Current ETA is mid November

FL33TW00D commented 10 months ago

@Mineru98 Large-v2 is comfortably supported on v0.8.0 now, file size is 1.7GB zstd.