Voice capture ends too quickly

[DevinloperNL]

First off, I’m really enjoying this package — thanks for the great work!

I have one concern around the capture lifecycle. Currently, the capture seems to end after a very short period of silence (roughly ~500ms). This makes it difficult to handle natural pauses during speech.

Would it be possible to support a continuous capture mode, where recording continues as long as start remains true, and only stops when it is explicitly set to false?

This would allow integrators to:

• Define their own silence timeout logic
• Provide an explicit “Stop” button
• Avoid relying on SpeechRecognitionResult.isFinal, which can be unreliable and browser-dependent

I think this would make the component significantly more flexible for real-world usage.

Thanks again!