Streamlit based Web App for ASR using Whisper Models

[prateekralhan]

Hi,

Kudos to the team for their work on ASR. I worked on developing a simple streamlit based web-app on automatic speech recognition for different audio formats using OpenAI's Whisper models 😄!

Title: Automatic Speech Recognition using OpenAI's Whisper
Github Repo: https://github.com/prateekralhan/OpenAI_Whisper_ASR

Happy opensourcing!

[Abdurrafey-Siddiqui]

Hi can you add options that allow us to tweak the transcribe like temperature, intitial_prompt ect..

[prateekralhan]

@Abdurrafey-Siddiqui currently I don't have the bandwidth to work on this. If you wish. you can raise a PR with these refinements else give me some time please to work on this.

[gustavz]

It would be nice to be able to record audio via browser. Unfortunately I did not find a proper solution for that using streamlit.
Using system audio of course works, but for a deployed app only client side audio makes sense.
Do you have a solution for that?

[prateekralhan]

@gustavz , can you have a look at this?

• https://github.com/whitphx/streamlit-webrtc

[gustavz]

yeah that's what I ended up using. Check it out the project: https://github.com/gustavz/jaivus

[prateekralhan]

This looks pretty cool. Great stuff!