谷歌语音到 C# 中的文本 API [英] Google speech to text API in C#
本文介绍了谷歌语音到 C# 中的文本 API的处理方法，对大家解决问题具有一定的参考价值，需要的朋友们下面随着小编来一起学习吧！
问题描述
当flac文件使用windows的录音机录制手册并使用软件转换器进行转换时，我的程序从谷歌得到了正确的响应.
但是当我使用我的程序记录的文件时，我得到了 "{"result":[]}来自谷歌.我该怎么办?这是我的代码:
发件人:
private static void CopyStream(FileStream fileStream, Stream requestStream){var 缓冲区 = 新字节 [32768];读入；while ((read = fileStream.Read(buffer, 0, buffer.Length)) > 0){requestStream.Write(buffer, 0, read);}}私有静态无效配置请求(HttpWebRequest请求){request.KeepAlive = true;request.SendChunked = true;request.ContentType = "audio/x-flac; rate=44100";request.UserAgent ="Mozilla/5.0 (Windows NT 6.2; WOW64) AppleWebKit/535.2 (KHTML, like Gecko) Chrome/15.0.874.121 Safari/535.2";request.Headers.Set(HttpRequestHeader.AcceptEncoding, "gzip,deflate,sdch");request.Headers.Set(HttpRequestHeader.AcceptLanguage, "en-GB,en-US;q=0.8,en;q=0.6");request.Headers.Set(HttpRequestHeader.AcceptCharset, "ISO-8859-1,utf-8;q=0.7,*;q=0.3");request.Method = "POST";}使用 (var fileStream = new FileStream(@"C:UsersAhmad MustofaDocumentsVisual Studio 2010ProjectsFPFPinDebugvoice.flac", FileMode.Open)){const string requestUrl = "https://www.google.com/speech-api/v2/recognize?output=json&lang=ar-sa&key=AIzaSyBJ6VJ326Rpb23msih2wGhXENEwU1TF1PA&client=chromium&maxresults=1&p;p;var request = (HttpWebRequest)WebRequest.Create(requestUrl);配置请求(请求)；var requestStream = request.GetRequestStream();CopyStream(fileStream, requestStream);使用 (var response = request.GetResponse()){使用 (var responseStream = response.GetResponseStream()){使用 (var zippedStream = new GZipStream(responseStream, CompressionMode.Decompress)){使用 (var sr = new StreamReader(zippedStream)){var res = sr.ReadToEnd();state.Text = res;}}}}}wav 录音机:
private void sourceStream_DataAvailable(object sender, NAudio.Wave.WaveInEventArgs e){如果(waveWriter == null)返回；waveWriter.WriteData(e.Buffer, 0, e.BytesRecorded);waveWriter.Flush();}fileName = "C:\Users\Ahmad Mustofa\Documents\Visual Studio 2010\Projects\FP\FP\bin\debug\voice.wav";int deviceNumber = hardware.SelectedItems[0].Index;尝试{sourceStream = new NAudio.Wave.WaveIn();sourceStream.DeviceNumber = deviceNumber;sourceStream.WaveFormat = new NAudio.Wave.WaveFormat(44100, NAudio.Wave.WaveIn.GetCapabilities(deviceNumber).Channels);sourceStream.DataAvailable += new EventHandler(sourceStream_DataAvailable);waveWriter = new NAudio.Wave.WaveFileWriter(fileName, sourceStream.WaveFormat);sourceStream.StartRecording();}捕获(异常前){state.Text = "disini" + ex.Message;} flac 转换器:
string inputFile = Path.Combine("wav", input);string outputFile = Path.Combine("flac", Path.ChangeExtension(input, ".flac"));如果 (!File.Exists(inputFile))throw new ApplicationException("找不到输入文件" + inputFile + "！");WavReader wav = new WavReader(inputFile);使用 (var flacStream = File.Create(outputFile)){FlacWriter flac = new FlacWriter(flacStream, wav.BitDepth, wav.Channels, wav.SampleRate);//缓冲 1 秒的音频数据byte[] 缓冲区 = 新字节 [wav.Bitrate/8];int 字节读取；做{bytesRead = wav.InputStream.Read(buffer, 0, buffer.Length);flac.Convert(buffer, 0, bytesRead);} while (bytesRead > 0);flac.Dispose();flac = 空；} 解决方案
谷歌云 api url 中明确提到，即
https:///cloud.google.com/speech-to-text/docs/async-recognize#speech-async-recognize-gcs-protocol
如果操作还没有完成，你可以通过重复发出GET请求来轮询端点，直到响应的done属性为真.
<代码> {"name": "此处的操作名称",元数据":{"@type": "type.googleapis.com/google.cloud.speech.v1.LongRunningRecognizeMetadata",进度百分比":0，"startTime": "2018-12-18T10:56:09.425584Z","lastUpdateTime": "2018-12-18T11:10:27.147310Z"},完成":真的，}通过重复发出 GET 请求来轮询端点，直到响应的 done 属性为真，或者您可以检查 "progressPercent": 0 直到它的值变为 100.一旦它的 100% 则 done 属性变为真.
我在我的代码中使用操作名称做了同样的事情，这里是代码供参考
public async TaskTranscribeLongMediaFile(字符串操作名称){string bearerToken = GetOAuthToken();var baseUrl = new Uri(googleSpeechBaseUrl + operationName);字符串 resultContent = string.Empty;使用 (var client = new HttpClient()){client.DefaultRequestHeaders.Add(HttpRequestHeader.Authorization.ToString(), "Bearer " + bearerToken);client.DefaultRequestHeaders.Add(HttpRequestHeader.ContentType.ToString(), "application/json; charset=utf-8");client.Timeout = TimeSpan.FromMilliseconds(Timeout.Infinite);int currentPercentage = 0;bool responseStatus = false;而(！响应状态){响应状态 = 假；//发送请求使用 (var 结果 = await client.GetAsync(baseUrl)){resultContent = 等待 result.Content.ReadAsStringAsync();ResponseObject responseObject = JsonConvert.DeserializeObject(resultContent);currentPercentage = responseObject.metadata.progressPercent;responseStatus = (responseObject.done && currentPercentage == 100);//根据百分比值延迟请求重复发出 GET 请求，直到响应的 done 属性为真.等待 Task.Delay(CalculateDealy(currentPercentage));}}};返回结果内容；} 为了延迟get请求:
////<摘要>///将请求延迟到毫秒数///</总结>///<param name="currentPercentage"></param>///<returns></returns>私人 int CalculateDealy(int currentPercentage){int x = currentPercentage/10;返回 (10 - x) * 1500；}获取身份验证令牌:
////<摘要>///获取 OAuth 令牌///</总结>///<returns></returns>公共字符串 GetOAuthToken(){返回 googleCredential.UnderlyingCredential.GetAccessTokenForRequestAsync("https://accounts.google.com/o/oauth2/v2/auth", CancellationToken.None).Result;}最后你会得到这样的结果:
<代码> {"name": "此处的操作名称",元数据":{"@type": "type.googleapis.com/google.cloud.speech.v1.LongRunningRecognizeMetadata",进度百分比":100，"startTime": "2018-12-18T10:56:09.425584Z","lastUpdateTime": "2018-12-18T11:10:27.147310Z"},完成":真的，回复": {"@type": "type.googleapis.com/google.cloud.speech.v1.LongRunningRecognizeResponse",结果": [{备择方案": [{"transcript": "好的，让我们开始吧",信心":0.97442055}]}， 等等 .....需要的东西:
- api-key.json 文件
- 安装包 Google.Apis.Auth.OAuth2 以授权HTTP 网络请求
谢谢
My program get a correct respon from google when the flac file recorded manual by using windows's sound recorder and convert it using a software converter.
But when I use the file that recorded by my program, I got "{"result":[]}
" from google. What should I do?
here is my code :
the sender :
private static void CopyStream(FileStream fileStream, Stream requestStream)
{
var buffer = new byte[32768];
int read;
while ((read = fileStream.Read(buffer, 0, buffer.Length)) > 0)
{
requestStream.Write(buffer, 0, read);
}
}
private static void ConfigureRequest(HttpWebRequest request)
{
request.KeepAlive = true;
request.SendChunked = true;
request.ContentType = "audio/x-flac; rate=44100";
request.UserAgent =
"Mozilla/5.0 (Windows NT 6.2; WOW64) AppleWebKit/535.2 (KHTML, like Gecko) Chrome/15.0.874.121 Safari/535.2";
request.Headers.Set(HttpRequestHeader.AcceptEncoding, "gzip,deflate,sdch");
request.Headers.Set(HttpRequestHeader.AcceptLanguage, "en-GB,en-US;q=0.8,en;q=0.6");
request.Headers.Set(HttpRequestHeader.AcceptCharset, "ISO-8859-1,utf-8;q=0.7,*;q=0.3");
request.Method = "POST";
}
using (var fileStream = new FileStream(@"C:UsersAhmad MustofaDocumentsVisual Studio 2010ProjectsFPFPinDebugvoice.flac", FileMode.Open))
{
const string requestUrl = "https://www.google.com/speech-api/v2/recognize?output=json&lang=ar-sa&key=AIzaSyBJ6VJ326Rpb23msih2wGhXENEwU1TF1PA&client=chromium&maxresults=1&pfilter=2";
var request = (HttpWebRequest)WebRequest.Create(requestUrl);
ConfigureRequest(request);
var requestStream = request.GetRequestStream();
CopyStream(fileStream, requestStream);
using (var response = request.GetResponse())
{
using (var responseStream = response.GetResponseStream())
{
using (var zippedStream = new GZipStream(responseStream, CompressionMode.Decompress))
{
using (var sr = new StreamReader(zippedStream))
{
var res = sr.ReadToEnd();
state.Text = res;
}
}
}
}
}
the wav recorder:
private void sourceStream_DataAvailable(object sender, NAudio.Wave.WaveInEventArgs e)
{
if (waveWriter == null) return;
waveWriter.WriteData(e.Buffer, 0, e.BytesRecorded);
waveWriter.Flush();
}
fileName = "C:\Users\Ahmad Mustofa\Documents\Visual Studio 2010\Projects\FP\FP\bin\debug\voice.wav";
int deviceNumber = hardware.SelectedItems[0].Index;
try
{
sourceStream = new NAudio.Wave.WaveIn();
sourceStream.DeviceNumber = deviceNumber;
sourceStream.WaveFormat = new NAudio.Wave.WaveFormat(44100, NAudio.Wave.WaveIn.GetCapabilities(deviceNumber).Channels);
sourceStream.DataAvailable += new EventHandler<NAudio.Wave.WaveInEventArgs>(sourceStream_DataAvailable);
waveWriter = new NAudio.Wave.WaveFileWriter(fileName, sourceStream.WaveFormat);
sourceStream.StartRecording();
}
catch (Exception ex)
{
state.Text = "disini" + ex.Message;
}
flac converter:
string inputFile = Path.Combine("wav ", input);
string outputFile = Path.Combine("flac", Path.ChangeExtension(input, ".flac"));
if (!File.Exists(inputFile))
throw new ApplicationException("Input file " + inputFile + " cannot be found!");
WavReader wav = new WavReader(inputFile);
using (var flacStream = File.Create(outputFile))
{
FlacWriter flac = new FlacWriter(flacStream, wav.BitDepth, wav.Channels, wav.SampleRate);
// Buffer for 1 second's worth of audio data
byte[] buffer = new byte[wav.Bitrate / 8];
int bytesRead;
do
{
bytesRead = wav.InputStream.Read(buffer, 0, buffer.Length);
flac.Convert(buffer, 0, bytesRead);
} while (bytesRead > 0);
flac.Dispose();
flac = null;
}
解决方案
It's clearly mentioned in the Google cloud api url i.e
https://cloud.google.com/speech-to-text/docs/async-recognize#speech-async-recognize-gcs-protocol
If the operation has not completed, you can poll the endpoint by repeatedly making the GET request until the done property of the response is true.
{
"name": "operationname here",
"metadata": {
"@type": "type.googleapis.com/google.cloud.speech.v1.LongRunningRecognizeMetadata",
"progressPercent": 0,
"startTime": "2018-12-18T10:56:09.425584Z",
"lastUpdateTime": "2018-12-18T11:10:27.147310Z"
},
"done": true,
}
poll the endpoint by repeatedly making the GET request until the done property of the response is true or you can check for the "progressPercent": 0 until it's value become 100. Once its 100 percent then done property becomes true.
I did the same in my code using operation name, for reference here is the code
public async Task<string> TranscribeLongMediaFile(string operationName)
{
string bearerToken = GetOAuthToken();
var baseUrl = new Uri(googleSpeechBaseUrl + operationName);
string resultContent = string.Empty;
using (var client = new HttpClient())
{
client.DefaultRequestHeaders.Add(HttpRequestHeader.Authorization.ToString(), "Bearer " + bearerToken);
client.DefaultRequestHeaders.Add(HttpRequestHeader.ContentType.ToString(), "application/json; charset=utf-8");
client.Timeout = TimeSpan.FromMilliseconds(Timeout.Infinite);
int currentPercentage = 0;
bool responseStatus = false;
while (!responseStatus)
{
responseStatus = false;
// Send request
using (var result = await client.GetAsync(baseUrl))
{
resultContent = await result.Content.ReadAsStringAsync();
ResponseObject responseObject = JsonConvert.DeserializeObject<ResponseObject>(resultContent);
currentPercentage = responseObject.metadata.progressPercent;
responseStatus = (responseObject.done && currentPercentage == 100);
// Delay the request based on percentage value to repeatedly making the GET request until the done property of the response is true.
await Task.Delay(CalculateDealy(currentPercentage));
}
}
};
return resultContent;
}
In order to delay the get request:
/// <summary>
/// Delay the request to number of milliseconds
/// </summary>
/// <param name="currentPercentage"></param>
/// <returns></returns>
private int CalculateDealy(int currentPercentage)
{
int x = currentPercentage / 10;
return (10 - x) * 1500;
}
Get auth token:
/// <summary>
/// Get OAuth token
/// </summary>
/// <returns></returns>
public string GetOAuthToken()
{
return googleCredential.UnderlyingCredential.GetAccessTokenForRequestAsync("https://accounts.google.com/o/oauth2/v2/auth", CancellationToken.None).Result;
}
At last, you will get the result like:
{
"name": "operationname here",
"metadata": {
"@type": "type.googleapis.com/google.cloud.speech.v1.LongRunningRecognizeMetadata",
"progressPercent": 100,
"startTime": "2018-12-18T10:56:09.425584Z",
"lastUpdateTime": "2018-12-18T11:10:27.147310Z"
},
"done": true,
"response": {
"@type": "type.googleapis.com/google.cloud.speech.v1.LongRunningRecognizeResponse",
"results": [
{
"alternatives": [
{
"transcript": "okay let's get started",
"confidence": 0.97442055
}
]
}, and so on .....
Things required:
- api-key.json file
- Install package Google.Apis.Auth.OAuth2 in order to authorize the HTTP web request
Thanks
这篇关于谷歌语音到 C# 中的文本 API的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！
查看全文
