Expand Minimize Picture-in-picture Power Device Status Voice Recognition Skip Back Skip Forward Minus Plus Play Search
Documentation
Getting Microphone Audio

Getting Microphone Audio

Capturing in-car audio allows developers to interact with users by requesting raw audio data provided to them from the car's microphones. In order to gather the raw audio from the vehicle, you must leverage the PerformAudioPassThru RPC.

SDL does not support automatic speech cancellation detection, so if this feature is desired, it is up to the developer to implement. The user may press an "OK" or "Cancel" button, the dialog may timeout, or you may close the dialog with EndAudioPassThru.

Note

SDL does not support an open microphone. However, SDL is working on wake-word support in the future. You may implement a voice command and start an audio pass thru session when that voice command occurs.

Starting Audio Capture

Before you start an audio capture session you need to find out what audio pass thru capabilities the module supports. You can then use that information to start an audio pass thru session.

Getting the Supported Capabilities

You must use a sampling rate, bit rate, and audio type supported by the module. Once you have successfully connected to the module, you can access these properties on the sdlManager.getSystemCapabilityManager instance.

sdlManager.getSystemCapabilityManager().getCapability(SystemCapabilityType.AUDIO_PASSTHROUGH, new OnSystemCapabilityListener() {
    @Override
    public void onCapabilityRetrieved(Object capability) {
        List<AudioPassThruCapabilities> audioPassThruCapabilities = (List<AudioPassThruCapabilities>) capability;
    }

    @Override
    public void onError(String info) {
        // Handle Error
    }
}, false);

The module may return one or multiple supported audio pass thru capabilities. Each capability will have the following properties:

Audio Pass Thru Capability Parameter Name Description
Sampling Rate samplingRate The sampling rate
Bits Per Sample bitsPerSample The sample depth in bits
Audio Type audioType The audio type

Sending the Audio Capture Request

To initiate audio capture, first construct a PerformAudioPassThru request.

PerformAudioPassThru audioPassThru = new PerformAudioPassThru();
audioPassThru.setAudioPassThruDisplayText1("Ask me \"What's the weather?\"");
audioPassThru.setAudioPassThruDisplayText2("or \"What's 1 + 2?\"");

TTSChunk initialPrompt = new TTSChunk("Ask me What's the weather? or What's 1 plus 2?", SpeechCapabilities.TEXT);
audioPassThru.setInitialPrompt(Arrays.asList(initialPrompt));
audioPassThru.setSamplingRate(SamplingRate._22KHZ);
audioPassThru.setMaxDuration(7000);
audioPassThru.setBitsPerSample(BitsPerSample._16_BIT);
audioPassThru.setAudioType(AudioType.PCM);
audioPassThru.setMuteAudio(false);

audioPassThru.setOnRPCResponseListener(new OnRPCResponseListener() {
    @Override
    public void onResponse (int correlationId, RPCResponse response) {
        switch (response.getResultCode()) {
            case SUCCESS:
                // The audio pass thru ended successfully. Process the audio data
            case ABORTED:
                // The audio pass thru was aborted by the user. You should cancel any usage of the audio data.
            default:
                // Some other error occurred. Handle the error.
        }
    }

    @Override
    public void onError (int correlationId, Result resultCode, String info) {
        // Handle error
    }
});

sdlManager.sendRPC(audioPassThru);

Ford Audio Pass Thru

Gathering Audio Data

SDL provides audio data as fast as it can gather it and sends it to the developer in chunks. In order to retrieve this audio data, the developer must observe the OnAudioPassThru notification.

Note

This audio data is only the current chunk of audio data, so the app is in charge of saving previously retrieved audio data.

sdlManager.addOnRPCNotificationListener(FunctionID.ON_AUDIO_PASS_THRU, new OnRPCNotificationListener() {
    @Override
    public void onNotified(RPCNotification notification) {
        OnAudioPassThru onAudioPassThru = (OnAudioPassThru) notification;
        byte[] dataRcvd = onAudioPassThru.getAPTData();
        // Do something with current audio data
    }
});

Format of Audio Data

The format of audio data is described as follows:

  • It does not include a header (such as a RIFF header) at the beginning.
  • The audio sample is in linear PCM format.
  • The audio data includes only one channel (i.e. monaural).
  • For bit rates of 8 bits, the audio samples are unsigned. For bit rates of 16 bits, the audio samples are signed and are in little-endian.

Ending Audio Capture

PerformAudioPassThru is a request that works in a different way than other RPCs. For most RPCs, a request is followed by an immediate response, with whether that RPC was successful or not. This RPC, however, will only send out the response when the audio pass thru has ended.

Audio capture can be ended four ways:

  1. The audio pass thru has timed out.
    • If the audio pass thru surpasses the timeout duration, this request will be ended with a resultCode of SUCCESS. You should handle the audio pass thru as though it was successful.
  2. The audio pass thru was closed due to user pressing "Cancel" (or other head-unit provided cancellation button).
    • If the audio pass thru was displayed, and the user pressed the "Cancel" button, you will receive a resultCode of ABORTED. You should ignore the audio pass thru.
  3. The audio pass thru was closed due to user pressing "Done" (or other head-unit provided completion button).
    • If the audio pass thru was displayed and the user pressed the "Done" button, you will receive a resultCode of SUCCESS. You should handle the audio pass thru as though it was successful.
  4. The audio pass thru was ended due to a request from the app for it to end.
    • If the audio pass thru was displayed, but you have established on your own that you no longer need to capture audio data, you can send an EndAudioPassThru RPC. You will receive a resultCode of SUCCESS. Depending on the reason that you sent the EndAudioPassThru RPC, you can choose whether or not to handle the audio pass thru as though it were successful. See Manually Stopping Audio Capture below for more details.

Manually Stopping Audio Capture

To force stop audio capture, simply send an EndAudioPassThru request. Your PerformAudioPassThru request will receive response with a resultCode of SUCCESS when the audio pass thru has ended.

EndAudioPassThru endAudioPassThru = new EndAudioPassThru();
endAudioPassThru.setOnRPCResponseListener(new OnRPCResponseListener() {
    @Override
    public void onResponse (int correlationId, RPCResponse response) {
        if (!response.getSuccess())) {
            // There was an error sending the end audio pass thru
            return;
        }

        // The end audio pass thru was sent successfully
    }
});

sdlManager.sendRPC(endAudioPassThru);

Handling the Response

To process the response received from an ended audio capture, make sure that you are listening to the PerformAudioPassThru response. If the response has a successful result, all of the audio data for the audio pass thru has been received and is ready for processing.

View on GitHub.com
Previous Section Next Section