Show HN: OWhisper – Ollama for realtime speech-to-text
docs.hyprnote.comHello everyone. This is Yujong from the Hyprnote team (https://github.com/fastrepl/hyprnote).
We built OWhisper for 2 reasons: (Also outlined in https://docs.hyprnote.com/owhisper/what-is-this)
(1). While working with on-device, realtime speech-to-text, we found there isn't tooling that exists to download / run the model in a practical way.
(2). Also, we got frequent requests to provide a way to plug in custom STT endpoints to the Hyprnote desktop app, just like doing it with OpenAI-compatible LLM endpoints.
The (2) part is still kind of WIP, but we spent some time writing docs so you'll get a good idea of what it will look like if you skim through them.
For (1) - You can try it now. (https://docs.hyprnote.com/owhisper/cli/get-started)
bash
brew tap fastrepl/hyprnote && brew install owhisper
owhisper pull whisper-cpp-base-q8-en
owhisper run whisper-cpp-base-q8-en
We're here and looking forward to your comments!
Thank you for taking the time to build something and share it. However what is the advantage of using this over whisper.cpp stream that can also do real time conversion?
https://github.com/ggml-org/whisper.cpp/tree/master/examples...
Ok, cool! I was actually one of the people on the hyprnote HN thread asking for a headless mode!
I was actually integrating some whisper tools yesterday. I was wondering if there was a way to get a streaming response, and was thinking it'd be nice if you can.
I'm on linux, so don't think I can test out owhisper right now, but is that a thing that's possible?
Also, it looks like the `owhisper run` command gives it's output as a tui. Is there an option for a plain text response so that we can just pipe it to other programs? (maybe just `kill`/`CTRL+C` to stop the recording and finalize the words).
Same question for streaming, is there a way to get a streaming text output from owhisper? (it looks like you said you create a deepgram compatible api, I had a quick look at the api docs, but I don't know how easy it is to hook into it and get some nice streaming text while speaking).
Oh yeah, and diarisation (available with a flag?) would be awesome, one of the things that's missing from most of the easiest to run things I can find.
> I'm on linux
I didn't tested on Linux yet, but we have linux build: http://owhisper.hyprnote.com/download/latest/linux-x86_64
> also, it looks like the `owhisper run` command gives it's output as a tui. Is there an option for a plain tex
`owhisper run` is more like way to quickly trying it out. But I think piping is definitely something that should work.
> Same question for streaming, is there a way to get a streaming text output from owhisper?
You can use Deepgram client to talk to `owhisper serve`. (https://docs.hyprnote.com/owhisper/deepgram-compatibility) So best resource might be Deepgram client SDK docs.
> diarisation
yeah on the roadmap
Nice stuff, had a quick test on linux and it works (built directly, I didn't check out the brew). I ran into a small issue with moonshine and opened an issue on github.
Great work on this! excited to keep an eye on things.
Oh wait, maybe you do support linux for owhisper: https://github.com/fastrepl/homebrew-hyprnote/blob/main/Form...
Can you help me out to find where the code you've built is? I can see the folder in github[0], but I can't see the code for the cli for instance? unless I'm blind.
[0] https://github.com/fastrepl/hyprnote/tree/main/owhisper
This is CLI entry point:
https://github.com/fastrepl/hyprnote/blob/8bc7a5eeae0fe58625...
Please find a way to add speaker diarization, with a way to remember the speakers. You can do it with pyannote, and get a vector embedding of each speaker that can be compared between audio samples, but that’s a year old now so I’m sure there’s better options now!
yeah that is on the roadmap!
Wait, this is cool.
I just spent last week researching the options (especially for my M1!) and was left wishing for a standard, full-service (live) transcription server for Whisper like OLlama has been for LLMs.
I’m excited to try this out and see your API (there seems to be a standard vaccuum here due to openai not having a real time transcription service, which I find to be a bummer)!
Edit: They seem to emulate the Deepgram API (https://developers.deepgram.com/reference/speech-to-text-api...), which seems like a solid choice. I’d definitely like to see a standard emerging here.
Correct. About the deepgram-compatibility: https://docs.hyprnote.com/owhisper/deepgram-compatibility
Let me know how it goes!
I’d like to use this to transcribe meeting minutes with multiple people. How could this program work for that use case?
If your use-case is meeting, https://github.com/fastrepl/hyprnote is for you. OWhisper is more like a headless version of it.
Can you describe how it pick different voices? Does it need separate audio channels, or does it recognize different voices on the same audio input?
It separate mic/speaker as 2 channel. So you can reliably get "what you said" vs "what you heard".
For splitting speaker within channel, we need AI model to do that. It is not implemented yet, but I think we'll be in good shape somewhere in September.
Also we have transcript editor that you can easily split segment, assign speakers.
If you want to transcribe meeting notes, whisper isn't the best tool because it doesn't separate the transcribe by speakers. There are some other tools that do that, but I'm not sure what the best local option is. I've used Google's cloud STT with the diarization option and manually renamed "Speaker N" after the fact.
Happy to answer any questions!
These are list of local models it supports:
- whisper-cpp-base-q8
- whisper-cpp-base-q8-en
- whisper-cpp-tiny-q8
- whisper-cpp-tiny-q8-en
- whisper-cpp-small-q8
- whisper-cpp-small-q8-en
- whisper-cpp-large-turbo-q8
- moonshine-onnx-tiny
- moonshine-onnx-tiny-q4
- moonshine-onnx-tiny-q8
- moonshine-onnx-base
- moonshine-onnx-base-q4
- moonshine-onnx-base-q8
FYI: owhisper pull whisper-cpp-large-turbo-q8 Failed to download model.ggml: Other error: Server does not support range requests. Got status: 200 OK
But the base-q8 works (and works quite well!). The TUI is really nice. Speaker diarization would make it almost perfect for me. Thanks for building this.
Sorry, maybe I missed it but I didn't see this list on your website. I think it is a good idea to add this info there. Besides that, thank you for the effort and your work! I will definetely give it a try
got it. fyi if you run `owhisper pull --help`, this info is printed
I thought whisper and others took large chunks (20-30 seconds) of speech, or a complete wave file as input. How do you get real-time transcription? What size chunks do you feed it?
To me, STT should take a continuous audio stream and output a continuous text stream.
I use VAD to chunk audio.
Whisper and Moonshine both works in a chunk, but for moonshine:
> Moonshine's compute requirements scale with the length of input audio. This means that shorter input audio is processed faster, unlike existing Whisper models that process everything as 30-second chunks. To give you an idea of the benefits: Moonshine processes 10-second audio segments 5x faster than Whisper while maintaining the same (or better!) WER.
Also for kyutai, we can input continuous audio in and get continuous text out.
- https://github.com/moonshine-ai/moonshine - https://docs.hyprnote.com/owhisper/configuration/providers/k...
Something like that, in a cli tool, that just gives text to stdout would be perfect for a lot of use cases for me!
(maybe with an `owhisper serve` somewhere else to start the model running or whatever.)
Are you thinking about the realtime use-case or batch use-case?
For just transcribing file/audio,
`owhisper run <MODEL> --file a.wav` or
`curl httpsL//something.com/audio.wav | owhisper run <MODEL>`
might makes sense.
agreed, both of those make sense, but I was thinking realtime. (pipes can stream data, I'd like and find useful something that can stream tts to stdout in realtime.)
It's open-source. Happy to review & merge if you can send us PR!
https://github.com/fastrepl/hyprnote/blob/8bc7a5eeae0fe58625...
I suggest you don't brand this "Ollama for X". They've become a commercial operation that is trying to FOSS-wash their actions through using llama.cpp's code and then throwing their users under the bus when they can't support them.
I see that you are also using llama.cpp's code? That's cool, but make sure you become a member of that community, not an abuser.
yeah we use whisper.cpp for whisper inference. this is more like a community-focused project, not a commercial product!