Feature: Configurable timeout and retry logic for transcription services

[armoucar-neon]

Problem

When using Whispering (I use OpenAi models), occasional connection delays can cause the application to hang for an extended period (~1 minute or more, but usually I have very fast transcription) while waiting for the API to respond. Currently, users have no control over:

• How long to wait before timing out
• User feedback during the pending request

Requested Features

• Configurable timeout duration for transcription API calls (per-model or global setting)
• User feedback during pending requests (progress indicator, timeout warning, or notification)
• Graceful degradation options (e.g., fallback to alternate transcription method if available)

Use Case

Users experience occasional latency from external transcription services. A configurable timeout would allow users to:

• Set appropriate wait times based on their network conditions
• Automatically retry failed requests without manual intervention
• Maintain control over the application experience during service delays

Scope

This should apply to all transcription services available in Whispering, not just a specific provider.

[Utkarsh9571]

i would like to work on that.

[armoucar-neon]

This is happening a lot. Most of my transcriptions execute in 3~5s but sometimes (quite often) it takes more than 1min and I have to keep waiting because I wouldn't like to repeat or type the whole prompt again. If I could set a 10s timeout with retries, I think that would improve the workflow.

[Utkarsh9571]

@armoucar-neon ok, i'll look into it.