Real-Time Recording and Call Detection

[DrCoffey]

DeepSqueak currently detects calls in completed audio recordings. However, real-time spectrogram visualization and call detection would be valuable.

• Allow DeepSqueak to record a real-time signal from an external microphone, soundcard, etc.
• Display real-time spectrogram within DeepSqueak App
• Give option to apply detection network to real-time recording
• Display detected calls in the DeepSqueak App
• Develop a method for triggering a hardware output when calls are detected

[VoiceScientist]

Huge fan of this task! Could I add independent multichannel monitoring and hardware triggers to the feature list? Let me know how my lab can help.

[DrCoffey]

Hey @VoiceScientist

Can you elaborate a little on your request here: https://gitter.im/DeepSqueak_Community/FeatureRequests

[vijayiyer05]

@DrCoffey, capturing some ideas from our brainstorming today:

• Integrating the Audio Toolbox is likely a top choice given its focus on streaming
• The implementation may not need (or be easily set up) to use a callback structure as in some other I/O streaming patterns from the past. An emerging pattern is to use parallel processing explicitly, e.g. to ensure apps stay responsive.
• For the graphics display, if not using it already, the animated line may be useful to minimize the time for graphics updates while streaming