Azure AI Speech でリアルタイム音声認識を行うコード

RealtimeAudioRecognitionByAzure.cs

// dotnet add package Microsoft.CognitiveServices.Speech --version 1.33.0
// dotnet add package NAudio --version 2.2.1
using Microsoft.CognitiveServices.Speech;
using Microsoft.CognitiveServices.Speech.Audio;
using NAudio.CoreAudioApi;

namespace RealtimeAudioRecognitionByAzure
{
    class Program
    {
        static async Task Main()
        {
            // AI Speech Service の設定を行う
            const string key = "";
            const string region = "japaneast";

            // オーディオデバイスを取得
            var enumerator = new MMDeviceEnumerator();
            var deviceType = DataFlow.Render; // オーディオデバイス
            //var deviceType = DataFlow.Capture; // マイクデバイス
            var devices = enumerator.EnumerateAudioEndPoints(deviceType, DeviceState.Active).ToList();
            if(!devices.Any())
            {
                Console.WriteLine("Audio device is not found.");
                return;
            }

            // 利用可能なオーディオデバイスを表示
            foreach(var (d, i) in devices.Select((d, i) => (d, i)))
            {
                Console.WriteLine($"{i}: {d.FriendlyName}");
            }

            // ユーザに使用するオーディオを選択させる
            Console.WriteLine("Select audio device to record:");
            var inputed = Console.ReadKey();
            if(!int.TryParse(inputed.KeyChar.ToString(), out var inputedIndex))
            {
                Console.WriteLine("Invalid input.");
                return;
            }
            var device = devices[inputedIndex];

            // AI Speech Service に接続して音声認識を開始する
            var config = SpeechConfig.FromSubscription(key, region);
            //config.SetProperty("ConversationTranscriptionInRoomAndOnline", "true");
            //config.SetProperty("DifferentiateGuestSpeakers", "true");
            //config.SetServiceProperty("transcriptionMode", "RealTimeAndAsync", ServicePropertyChannel.UriQueryParameter);
            var input = deviceType == DataFlow.Render
                ? AudioConfig.FromSpeakerOutput(device.ID)
                : AudioConfig.FromMicrophoneInput(device.ID);
            var recognizer = new SpeechRecognizer(config, "ja-JP", input);

            // 音声認識のイベントを設定する
            recognizer.SessionStarted += (s, e) =>
            {
                Console.WriteLine("STARTING on {0}", e);
            };
            recognizer.SessionStopped += (s, e) =>
            {
                Console.WriteLine("ClOSING on {0}", e);
            };
            recognizer.Canceled += (s, e) =>
            {
                Console.WriteLine("ClOSING on {0}", e);
            };
            recognizer.Recognized += (s, e) =>
            {
                Console.WriteLine($"RECOGNIZED: Text={e.Result.Text}");
            };
            recognizer.Recognizing += (s, e) =>
            {
                Console.WriteLine($"RECOGNIZING: Text={e.Result.Text}");
            };

            // 音声認識を開始する
            await recognizer.StartContinuousRecognitionAsync().ConfigureAwait(false);

            // 音声認識を停止する
            Console.WriteLine("Press any key to stop");
            Console.ReadLine();
            await recognizer.StopContinuousRecognitionAsync().ConfigureAwait(false);
        }
    }
}