feat: add voice search (speech-to-text) support #95
Description
Define You:
- Contributor
Is your feature request related to a problem? Please describe.
Typing in the search bar can be slow or inconvenient for many users. Modern browsers provide voice search for faster input, but Peersky Browser currently lacks this feature. Users who prefer speaking instead of typing need a more accessible way to search.
Describe the solution you'd like...
Add a voice search feature by integrating a microphone icon inside the search bar.
When clicked:
-
The browser should start listening to the user’s voice.
-
Speech should be converted into text automatically (speech-to-text).
-
The recognised text should appear inside the search bar in real time.
-
Auto language detection should be supported so the user doesn’t need to choose a language manually.
-
The voice input should stop when clicked again or after a short silence.
Describe alternatives you've considered?
-
Using Chromium’s built-in Web Speech API (but it’s not fully offline and may not auto-detect language correctly).
-
Using cloud speech APIs (Google, Azure, etc.), but these require paid plans and internet access.
Approach to be followed (optional):
-
Add a mic icon inside the search bar UI.
-
Use Vosk (offline STT engine) for speech-to-text with a multi-language model for auto detection.
-
Start audio recording when the mic icon is clicked.
-
Send audio stream to Vosk for real-time transcription.
-
Insert transcription into the search bar input.
-
Handle stop events:
-
Click again
-
Silence timeout
- Display a visual indicator (e.g., mic glowing) while recording.
Additional context
-
Vosk is free and works offline.
-
Chromium is open-source, no API payments required.
-
This feature improves accessibility and user experience.
-
Helps users who prefer speaking over typing.