public TranscriptionOutput(string text, Boolean transcriptionSuccess, AudioSegment segment, User speaker = null)
{
Text = text;
StartOffset = (int)segment.StartOffset;
EndOffset = (int)segment.EndOffset;
TranscriptionSuccess = transcriptionSuccess;
Segment = segment;
Speaker = speaker;
}
private async Task RecognitionWithPullAudioStreamAsync()
{
var stopRecognition = new TaskCompletionSource <int>();
var entireAudio = Controller.FileSplitter.GetEntireAudio();
using var audioInput = AudioConfig.FromStreamInput(entireAudio.AudioStream);
// Creates a speech recognizer using audio stream input.
using var speech_recogniser = new SpeechRecognizer(Controller.SpeechConfig, audioInput);
// Subscribes to events. Subscription is important, otherwise recognition events aren't handled.
speech_recogniser.Recognizing += (s, e) =>
{
//
};
speech_recogniser.Recognized += (s, e) =>
{
string transcribedText = "";
Boolean resultAvailable = false;
Boolean success = false;
if (e.Result.Reason == ResultReason.RecognizedSpeech)
{
resultAvailable = true;
Console.WriteLine($"RECOGNIZED: Text = {e.Result.Text}\n");
transcribedText = e.Result.Text; //Write transcription text to result
success = true; //Set flag to indicate that transcription succeeded. //Reset error counter
}
else if (e.Result.Reason == ResultReason.NoMatch)
{
resultAvailable = true;
Console.WriteLine($">\tNOMATCH: Speech could not be recognized.");
transcribedText = $"NOMATCH: Speech could not be recognized."; //Write fail message to result
}
if (resultAvailable)
{
/* Start and end offsets in milliseconds from 0, which is beginning of audio. Note
* conversion from ticks to milliseconds.*/
long startOffset = e.Result.OffsetInTicks / 10000L;
long endOffset = startOffset + (long)e.Result.Duration.TotalMilliseconds;
//CRITICAL section. Add the result to transcriptionOutputs wrapped in a TranscriptionOutput object.
lock (_lockObj)
{
/*Split the audio based on start and end offset of the identified phrase. Note access to shared stream. */
AudioSegment segment = Controller.FileSplitter.SplitAudio((ulong)startOffset, (ulong)endOffset);
TranscriptionOutputs.Add(startOffset, new TranscriptionOutput(transcribedText, success, segment));
} //END CRITICAL section.
}
};
speech_recogniser.Canceled += (s, e) =>
{
Console.WriteLine($">\tCANCELED: Reason = {e.Reason}");
Console.WriteLine(">\tTranscription Complete.");
if (e.Reason == CancellationReason.Error)
{
Console.WriteLine($">\tCANCELED: ErrorCode={e.ErrorCode}");
Console.WriteLine($">\tCANCELED: ErrorDetails={e.ErrorDetails}");
Console.WriteLine($">\tCANCELED: Make Sure to Update Subscription Info");
}
stopRecognition.TrySetResult(0);
};
speech_recogniser.SessionStarted += (s, e) =>
{
//Console.WriteLine(">\tSession Started.");
};
speech_recogniser.SessionStopped += (s, e) =>
{
//Console.WriteLine(">\tSession Stopped.");
stopRecognition.TrySetResult(0);
};
// Starts continuous recognition. Uses StopContinuousRecognitionAsync() to stop recognition.
await speech_recogniser.StartContinuousRecognitionAsync().ConfigureAwait(false);
// Waits for completion.
// Use Task.WaitAny to keep the task rooted.
Task.WaitAny(new[] { stopRecognition.Task });
// Stops recognition.
await speech_recogniser.StopContinuousRecognitionAsync().ConfigureAwait(false);
}
