Auto-Transcription Trigger Without Audio Input
App Name: ArcSphere
Platform: Android
Issue Type: Functional / UI Logic
Severity: Medium (Interferes with manual text input and wastes processing time)
1. Description
When the microphone/voice input is activated, the app immediately enters a "Transcribing..." state and attempts to generate text even when there is no audible speech or background noise. This results in unintended text being entered into the query bar (e.g., "you" appearing automatically) or the app getting stuck in a processing loop.
2. Steps to Reproduce
- Open the ArcSphere app.
- Tap the Microphone icon in the "Ask Anything" input bar.
- Remain silent and do not provide any audio input.
- Observe the "Transcribing..." status message.
- Notice that text (like "you") is generated and populated in the text box automatically.
3. Expected Result
The app should wait for detectable speech before displaying the "Transcribing..." state. If no audio is detected within a certain timeframe, it should ideally time out or remain idle rather than generating "phantom" text.
4. Actual Result
The app triggers the transcription process instantly and populates the input field with incorrect or irrelevant words despite total silence.
Technical Details (from Recording)
- Trigger Element: Microphone button.
- Behavior observed: The "Transcribing..." spinner appears immediately after the recording is stopped/paused, even with no voice activity.
- Model Selected: GPT-5 Mini (as seen in the dropdown).