How to receive phone call in azure and stream it into intent recognition (azure cognitive services)

CarlErikLindberg
October 10, 2023
156 views
0 votes
2 Answers

I am trying to receive a phone call and stream it into cognitive services. How can I do this?

I tried:
Setup phone number in Azure and make Azure Event grid and setup a webhook to send the the event to my app.
However I cannot seem to get a stream out of the received object.
Code:

[AllowAnonymous]
[HttpPost]
public async Task<IActionResult> IncomingCall([FromBody] EventGridEvent[] events)
{
            int eventCount = 0;

            foreach (var eventGridEvent in events)
            {
                try
                {

                    switch (eventGridEvent.EventType)
                    {
                        case SystemEventNames.EventGridSubscriptionValidation:
                            {
                                var eventData = eventGridEvent.Data.ToObjectFromJson<SubscriptionValidationEventData>();
                                var responseData = new SubscriptionValidationResponse
                                { ValidationResponse = eventData.ValidationCode };
                                if (responseData.ValidationResponse != null)
                                { return Ok(responseData);}
                            }
                            break;
                        case "Microsoft.Communication.IncomingCall":

                  var _client = new CallAutomationClient("<ACS connection string>");
                            var eventData2 = eventGridEvent.Data.ToObjectFromJson<AcsIncomingCallEventData>();
                            string incomingCallContext = eventData2.IncomingCallContext;
                            string serverCallId = eventData2.ServerCallId;
                            var answerCallOptions = new AnswerCallOptions(incomingCallContext, new Uri("wss://... my url here}"));
                            answerCallOptions.OperationContext = "";
                            var call = await _client.AnswerCallAsync(answerCallOptions);

                            // How to get the stream from the call?

                           // then stream into Azure CLU
                           // some constants
                           var speechConfig = SpeechConfig.FromSubscription(speechKey, speechRegion);
                           speechConfig.SpeechRecognitionLanguage = "en-US";
                      speechConfig.SetProperty(PropertyId.Speech_SegmentationSilenceTimeoutMs, "2000");
using AudioConfig audioConfig = AudioConfig.FromStreamInput(audioStream);
                           using (var intentRecognizer = new IntentRecognizer(speechConfig, audioConfig))

Alternative I tried this.
But then I get some text, and how can I stream that text into CLU (IntentRecognizer)?

                            var callAutomationClient = new CallAutomationClient("<ACS connection string>");

                            var answerCallOptions = new AnswerCallOptions("<Incoming call context once call is connected>", new Uri("<https://sample-callback-uri>"))
                            {
                                AzureCognitiveServicesEndpointUrl = new Uri("https://sample-cognitive-service-resource.cognitiveservices.azure.com/") // for Speech-To-Text (choices)
                            };
                            var answerCallResult = await callAutomationClient.AnswerCallAsync(answerCallOptions);

Answers

Chosen as BEST ANSWER
- CarlErikLindberg
- October 10, 2023 at 6:36 am
- 0 votes
0
Found a solution here

How to convert PCM S16LE audio to MU-LAW/8000 using .NET (Windows/Mac/Linux)

C# Code is here: https://github.com/chambersandpartners/twilio-transcription-poc/blob/master/TwilioMediaStreams/Services/MuLawEncoder.cs

(Edit)

In your code, you have already set up the EventGrid subscription and are receiving events. To handle incoming call events, you can use the following code

switch (eventGridEvent.EventType)
{
    case SystemEventNames.EventGridSubscriptionValidation:
    {
        // Handle subscription validation if necessary
        var eventData = eventGridEvent.Data.ToObjectFromJson<SubscriptionValidationEventData>();
        var responseData = new SubscriptionValidationResponse
        {
            ValidationResponse = eventData.ValidationCode
        };
        if (responseData.ValidationResponse != null)
        {
            return Ok(responseData);
        }
    }
    break;
    case "Microsoft.Communication.IncomingCall":
    {
        var eventData2 = eventGridEvent.Data.ToObjectFromJson<AcsIncomingCallEventData>();
        string incomingCallContext = eventData2.IncomingCallContext;
        string serverCallId = eventData2.ServerCallId;

        // Answer the incoming call and get the call object
        var _client = new CallAutomationClient("<ACS connection string>");
        var answerCallOptions = new AnswerCallOptions(incomingCallContext, new Uri("wss://... my url here}"));
        answerCallOptions.OperationContext = "";
        var call = await _client.AnswerCallAsync(answerCallOptions);

        // Now, you need to capture the audio stream from the call and send it to Azure Cognitive Services for speech recognition.
        
        // Stream the audio to Azure Cognitive Services
        var speechConfig = SpeechConfig.FromSubscription(speechKey, speechRegion);
        speechConfig.SpeechRecognitionLanguage = "en-US";
        speechConfig.SetProperty(PropertyId.Speech_SegmentationSilenceTimeoutMs, "2000");

        // Create an AudioConfig from the call's audio stream
        using AudioConfig audioConfig = AudioConfig.FromStreamInput(call.AudioStream);

        // Create an IntentRecognizer to perform speech recognition
        using (var intentRecognizer = new IntentRecognizer(speechConfig, audioConfig))
        {
            // Start recognizing and processing the audio stream
            var result = await intentRecognizer.RecognizeOnceAsync();
            
            // You can now process the result or send it to Azure Cognitive Language Understanding (LUIS) for further intent recognition.
        }
    }
    break;
}