Build real time speech2text web apps using OpenAI's Whisper https://openai.com/blog/whisper/
Stars
830
Forks
145
Watchers
830
Open Issues
16
Overall repository health assessment
No package.json found
This might not be a Node.js project
Merge remote-tracking branch 'origin/frontend' into frontend
4497c42View on GitHubChanged message to statusMessage, removed redundant setNewMessage function
8977f61View on GitHubEnhanced codebase clarity and adhered to best practices by significantly reducing the reliance on plain strings. Implemented data type indicators in functions handling audio data, incorporated assertions for audio data types, and transformed ClientState into an enum for improved representation. Whisper model sizes are now exclusively utilized as enum elements. Additionally, introduced a Whisper model cache to enable swift transcription startup, eliminating the requirement for model reloading. Fixed bug where the client's Whisper model size list didn't match the server's supported model sizes list. Fixed bug where "No more clients allowed" error doesn't appear correctly in client.
c825d5cView on GitHubEnhanced audio pre-processing by implementing voice activity detection for selecting speech-containing audio chunks. Also, addressed end-of-stream behavior for sequential mode and cleanup behavior for disconnected clients. Tidied up some logging messages for better clarity.
33534ebView on GitHub