How to record microphone to more compressed format during WebRTC call on Android?

On the Android platform, WebRTC is a widely adopted framework for real-time communication. When converting microphone audio streams into compressed formats, we typically process the audio within the communication pipeline to enhance compression efficiency, reduce bandwidth consumption, and preserve audio quality as much as possible. Below are key steps and approaches I've used to address this challenge:

1. Selecting the appropriate audio encoding format

Choosing the right audio encoding format is critical. For WebRTC, Opus is an excellent choice due to its superior compression ratio and audio quality. Opus dynamically adjusts the bitrate based on network conditions, making it ideal for real-time communication scenarios.

2. Configuring WebRTC's audio processor

WebRTC offers extensive APIs for configuring audio processing. By setting parameters in WebRtcAudioRecord, you can adjust the sampling rate, bitrate, and other settings. For instance, lowering the bitrate directly reduces data usage, but it's essential to balance audio quality and file size.

3. Real-time audio stream processing

Implementing a custom audio processing module allows preprocessing audio data before encoding. Using algorithms like AAC or MP3 can further compress the audio stream. This requires deep knowledge of WebRTC's audio processing framework and integrating custom logic at appropriate stages.

4. Monitoring and Tuning

Continuous monitoring of audio quality and compression effectiveness is vital. Tools like WebRTC's RTCPeerConnection getStats API provide real-time call quality metrics. Adjust compression parameters based on this data to optimize both call quality and data efficiency.

5. Example

Suppose we're developing an Android video conferencing app using WebRTC. To minimize data usage, we compress the audio stream by selecting Opus with a 24kbps bitrate—a setting that maintains clear voice quality while significantly reducing data transmission. Here's the configuration:

java
peerConnection = factory.createPeerConnection(rtcConfig, pcObserver);

MediaConstraints audioConstraints = new MediaConstraints();
audioConstraints.mandatory.add(new MediaConstraints.KeyValuePair("opus/48000", "24"));
audioConstraints.optional.add(new MediaConstraints.KeyValuePair("VoiceActivityDetection", "true"));

AudioSource audioSource = factory.createAudioSource(audioConstraints);
AudioTrack localAudioTrack = factory.createAudioTrack("101", audioSource);

This setup uses Opus at 24kbps, reducing bandwidth needs without compromising audio quality significantly.

By applying this method, we can effectively leverage WebRTC for real-time communication on Android while compressing audio to adapt to varying network conditions. This is especially crucial for mobile applications, which often operate in unstable network environments.