static StreamingRecognitionConfig CreateStreamingRecognitionConfig(
uint sampleRate,
string audioEncoding,
uint countChannel,
uint maxAlternatives,
bool disableAutomaticPunctuation,
bool enableInterimResults,
bool singleUtterance
)
{
var streamingRecognitionConfig = new StreamingRecognitionConfig();
streamingRecognitionConfig.Config = CreateRecognitionConfig(
sampleRate,
audioEncoding,
countChannel,
maxAlternatives,
disableAutomaticPunctuation
);
streamingRecognitionConfig.InterimResultsConfig = new InterimResultsConfig
{
EnableInterimResults = enableInterimResults,
};
streamingRecognitionConfig.SingleUtterance = singleUtterance;
return(streamingRecognitionConfig);
}
/// <summary>
/// Construct an InfiniteSpeechStream without starting it.
/// </summary>
/// <param name="wasapiCapture">A WasapiCapture that must be in stopped state. It will be disposed with this stream.</param>
/// <param name="video">If use video enhanced model.</param>
/// <param name="sampleRate">The sample rate of data sent to google.</param>
public InfiniteSpeechStream([NotNull] WasapiCapture wasapiCapture,
bool video = false,
[NonNegativeValue] int sampleRate = DefaultReSampleRate,
[NotNull] string language = LanguageCodes.English.UnitedStates)
{
_wasapiCapture = wasapiCapture;
GoogleCloudSpeechClient = SpeechClient.Create();
if (_wasapiCapture.CaptureState != CaptureState.Stopped)
{
throw new ArgumentException("WasapiCapture should not in capture state.", nameof(wasapiCapture));
}
_reSamplerWaveProvider = new BufferedWaveProvider(wasapiCapture.WaveFormat)
{
ReadFully = false
};
_reSampler = new MediaFoundationResampler(_reSamplerWaveProvider, new WaveFormat(sampleRate, 16, 1));
_convertedBuffer = new byte[_reSampler.WaveFormat.AverageBytesPerSecond * 10];
bool supportEnhanced = language == LanguageCodes.English.UnitedStates;
string model = supportEnhanced ? (video ? "video" : "phone_call") : "default";
_recognitionConfig = new StreamingRecognitionConfig()
{
Config = new RecognitionConfig()
{
Encoding = RecognitionConfig.Types.AudioEncoding.Linear16,
SampleRateHertz = sampleRate,
LanguageCode = language,
UseEnhanced = supportEnhanced,
EnableAutomaticPunctuation = true,
Model = model
},
InterimResults = true,
};
}
static void gRPc_connect(string path)
{
// создаем канал для обмена сообщениями с сервером
// параметр - адрес сервера gRPC
// создаем клиента
// var client = new Greeter.GreeterClient(channel);
// var mp3 = new MP3File(@"C:\Users\IlinKS\source\repos\JWT_gRPC_console\JWT_gRPC_console\2020.10.12_21.15.41_79214430425_incoming_mixed_79098420960.mp3","testik");
// string path = @"C:\Users\IlinKS\source\repos\JWT_gRPC_console\JWT_gRPC_console\2020.10.12_21.15.41_79214430425_incoming_mixed_79098420960.mp3";
// string path = @"c:\Users\ilinks\source\repos\JWT_gRPC_console\JWT_gRPC_console\6650322009_89136872782_2020-01-24_08_24_12.wav";
var audio = TagLib.File.Create(path);
var vkc = new VoiceKitClient("bnZ6cWRvcWJncWNud2dqcGxtZ21ndXVvZXdjaWpueHk=m.rogencovfilbert", "YXlzdXBvaWhrdmZzZmtvYXZtb3plaHZqeGlrcGZ1d2c=");
// первый запрос
StreamingRecognitionConfig streaming_config = new StreamingRecognitionConfig();
streaming_config.Config = new RecognitionConfig();
streaming_config.Config.SampleRateHertz = (uint)audio.Properties.AudioSampleRate;
streaming_config.Config.NumChannels = (uint)audio.Properties.AudioChannels;
audio.Dispose();
// using FileStream fstream = File.Open(path, FileMode.Open);
switch (Path.GetExtension(path))
{
case ".mp3":
streaming_config.Config.Encoding = AudioEncoding.MpegAudio;
vkc.StreamingRecognize(streaming_config, File.Open(path, FileMode.Open)).GetAwaiter().GetResult();
break;
case ".wav":
streaming_config.Config.Encoding = AudioEncoding.Linear16;
vkc.StreamingRecognizeWAV(streaming_config, WavToPcmConvert(path)).GetAwaiter().GetResult();
break;
}
return;
}
/// <summary>
/// 邊錄音, 邊發到Google轉文字
/// </summary>
public Task <SpeechData> StartStream()
{
return(Task.Run(() =>
{
GCP.Share.Init();
var StreamingConfig = new StreamingRecognitionConfig()
{
Config = new RecognitionConfig()
{
Encoding = RecognitionConfig.Types.AudioEncoding.Linear16,
SampleRateHertz = 16000,
LanguageCode = Language,
},
InterimResults = true,
};
var Streaming = GCP.Share.Client.StreamingRecognize();
Streaming.WriteAsync(new StreamingRecognizeRequest {
StreamingConfig = StreamingConfig
}).Wait();
Microphone.Share.OnReceive = (args) =>
{
if (Streaming != null)
{
try
{
Streaming.WriteAsync(
new StreamingRecognizeRequest()
{
AudioContent = Google.Protobuf.ByteString.CopyFrom(args.Buffer, 0, args.BytesRecorded)
}).Wait();
}
catch (Exception ex)
{
throw ex;
}
}
};
var speechAnaysis = Task.Run <string>(async() =>
{
sb = new StringBuilder();
try
{
while (await Streaming.ResponseStream.MoveNext(default))
public async Task StreamingRecognizeWAV(StreamingRecognitionConfig config, Stream audioStream)
{
var streamingSTT = _clientSTT.StreamingRecognize(GetMetadataSTT());
var requestWithConfig = new StreamingRecognizeRequest
{
StreamingConfig = config
};
await streamingSTT.RequestStream.WriteAsync(requestWithConfig); //передаем параметры аудио
Task PrintResponsesTask = Task.Run(async() => // Пишем ответ в консоль
{
while (await streamingSTT.ResponseStream.MoveNext())
{
foreach (var result in streamingSTT.ResponseStream.Current.Results)
{
foreach (var alternative in result.RecognitionResult.Alternatives)
{
System.Console.WriteLine(alternative.Transcript);
text += alternative.Transcript;
}
}
}
});
var buffer = new byte[config.Config.SampleRateHertz / 10];
int bytesRead;
// audioStream.Position = 44;
while ((bytesRead = audioStream.Read(buffer, 0, buffer.Length)) > 0)
{
await streamingSTT.RequestStream.WriteAsync(
new StreamingRecognizeRequest
{
AudioContent = Google.Protobuf
.ByteString.CopyFrom(buffer, 0, bytesRead),
});
}
await streamingSTT.RequestStream.CompleteAsync();
await PrintResponsesTask;
}
public async Task StreamingRecognize(StreamingRecognitionConfig config, Stream audioStream)
{
var streamingSTT = _clientSTT.StreamingRecognize(GetMetadataSTT());
var requestWithConfig = new StreamingRecognizeRequest
{
StreamingConfig = config
};
await streamingSTT.RequestStream.WriteAsync(requestWithConfig);
Task PrintResponsesTask = Task.Run(async() =>
{
while (await streamingSTT.ResponseStream.MoveNext())
{
foreach (var result in streamingSTT.ResponseStream.Current.Results)
{
foreach (var alternative in result.RecognitionResult.Alternatives)
{
System.Console.WriteLine(alternative.Transcript);
}
}
}
});
var buffer = new byte[2 * 1024];
int bytesRead;
while ((bytesRead = audioStream.Read(buffer, 0, buffer.Length)) > 0)
{
await streamingSTT.RequestStream.WriteAsync(
new StreamingRecognizeRequest
{
AudioContent = Google.Protobuf
.ByteString.CopyFrom(buffer, 0, bytesRead),
});
}
await streamingSTT.RequestStream.CompleteAsync();
await PrintResponsesTask;
}
static void Main(string[] args)
{
// 証明書を作成
var credential = GoogleCredential.FromJson(File.ReadAllText("SpeechTest-4db378c087bb.json"));
credential = credential.CreateScoped("https://www.googleapis.com/auth/cloud-platform");
// サーバに接続するためのチャンネルを作成
var channel = new Channel("speech.googleapis.com:443", credential.ToChannelCredentials());
// Google Speech APIを利用するためのクライアントを作成
var client = new Speech.SpeechClient(channel);
// ストリーミングの設定
var streamingConfig = new StreamingRecognitionConfig
{
Config = new RecognitionConfig
{
SampleRate = 16000,
Encoding = RecognitionConfig.Types.AudioEncoding.Linear16,
LanguageCode = "ja-JP",
},
InterimResults = true,
SingleUtterance = false,
};
// ストリーミングを開始
using (var call = client.StreamingRecognize())
{
Console.WriteLine("-----------\nstart.\n");
// Cloud Speech APIからレスポンスが返ってきた時の挙動を設定
var responseReaderTask = Task.Run(async() =>
{
// MoveNext１回につきレスポンス１回分のデータがくる
while (await call.ResponseStream.MoveNext())
{
var note = call.ResponseStream.Current;
// データがあれば、認識結果を出力する
if (note.Results != null && note.Results.Count > 0 &&
note.Results[0].Alternatives.Count > 0)
{
Console.WriteLine("result: " + note.Results[0].Alternatives[0].Transcript);
}
}
});
// 最初の呼び出しを行う。最初は設定データだけを送る
var initialRequest = new StreamingRecognizeRequest
{
StreamingConfig = streamingConfig,
};
call.RequestStream.WriteAsync(initialRequest).Wait();
// 録音モデルの作成
IAudioRecorder recorder = new RecordModel();
// 録音モデルが音声データを吐いたら、それをすかさずサーバに送信する
recorder.RecordDataAvailabled += (sender, e) =>
{
if (e.Length > 0)
{
// WriteAsyncは一度に一回しか実行できないので非同期処理の時は特に注意
// ここではlockをかけて処理が重ならないようにしている
lock (recorder)
{
call.RequestStream.WriteAsync(new StreamingRecognizeRequest
{
AudioContent = RecognitionAudio.FromBytes(e.Buffer, 0, e.Length).Content,
}).Wait();
}
}
};
// 録音の開始
recorder.Start();
// Cloud Speech APIのストリーミングは1回60秒までなので、50秒まできたら打ち切る
var timer = new Timer(1000 * 50);
timer.Start();
// 50秒経過した時、実際に打ち切るコード
timer.Elapsed += async(sender, e) =>
{
recorder.Stop();
await call.RequestStream.CompleteAsync();
};
// 待機
responseReaderTask.Wait();
// ここに到達した時点で、APIの呼び出しが終了したということなので、タイマーを切る
timer.Dispose();
}
Console.WriteLine("\n-----------\nCompleted (Time out)");
Console.ReadKey();
}
/// <summary>
/// Asynchronously streams audio to the Google Cloud Speech server and receives results.
/// </summary>
static async Task StreamingRequest()
{
using (var call = m_Client.StreamingRecognize())
{
Task responseReaderTask = Task.Run(async() =>
{
while (await call.ResponseStream.MoveNext())
{
StreamingRecognizeResponse response = call.ResponseStream.Current;
var responseJSON = new JSONObject();
if (response.Error != null)
{
var errorJSON = new JSONObject();
errorJSON.AddField("code", response.Error.Code);
errorJSON.AddField("message", response.Error.Message);
responseJSON.AddField("error", errorJSON);
}
if (response.Results != null && response.Results.Count > 0)
{
var resultsJSON = new JSONObject();
foreach (var result in response.Results)
{
var resultJSON = new JSONObject();
if (result.Alternatives != null && result.Alternatives.Count > 0)
{
var alternativesJSON = new JSONObject();
foreach (var alternative in result.Alternatives)
{
var alternativeJSON = new JSONObject();
alternativeJSON.AddField("transcript", alternative.Transcript);
alternativeJSON.AddField("confidence", alternative.Confidence);
alternativesJSON.Add(alternativeJSON);
}
resultJSON.AddField("alternatives", alternativesJSON);
resultJSON.AddField("is_final", result.IsFinal);
resultJSON.AddField("stability", result.Stability);
}
resultsJSON.Add(resultJSON);
}
responseJSON.AddField("results", resultsJSON);
}
responseJSON.AddField("result_index", response.ResultIndex);
responseJSON.AddField("endpointer_type", response.EndpointerType.ToString());
Console.WriteLine("response: " + responseJSON);
}
});
// Send initial config request
var configRequest = new StreamingRecognizeRequest();
var streamingRecognitionConfig = new StreamingRecognitionConfig();
var recognitionConfig = new RecognitionConfig();
recognitionConfig.Encoding = RecognitionConfig.Types.AudioEncoding.LINEAR16;
recognitionConfig.SampleRate = 16000;
streamingRecognitionConfig.InterimResults = true;
streamingRecognitionConfig.SingleUtterance = false;
streamingRecognitionConfig.Config = recognitionConfig;
configRequest.StreamingConfig = streamingRecognitionConfig;
await call.RequestStream.WriteAsync(configRequest);
// Send audio chunks
Task sendChunksTask = SendChunks(call);
await sendChunksTask;
await call.RequestStream.CompleteAsync();
await responseReaderTask;
call.Dispose();
}
}
