public virtual Stream GetAudioStream(string url, string text, SpeechLocaleOptions locale, VoiceName voiceName, GenderOptions voiceType, AudioOutputFormatOptions outputFormat, string token)
{
// Construct HTTP request to get the logo
HttpWebRequest httpRequest = (HttpWebRequest)
WebRequest.Create(url);
httpRequest.Method = WebRequestMethods.Http.Post;
var headers = GetAudioHeaders(token, outputFormat);
httpRequest.Headers.Add("X-Microsoft-OutputFormat", outputFormat.MemberAttrValue());
httpRequest.Headers.Add("Authorization", $"Bearer {token}");
httpRequest.ContentType = "application/ssml+xml";
httpRequest.UserAgent = "TTSClient";
byte[] reqData = Encoding.UTF8.GetBytes(GetAudioContent(text, locale, voiceName, voiceType));
httpRequest.ContentLength = (long)reqData.Length;
Stream requestStream = httpRequest.GetRequestStream();
requestStream.Write(reqData, 0, reqData.Length);
requestStream.Close();
// Get back the HTTP response for web server
HttpWebResponse httpResponse = (HttpWebResponse)httpRequest.GetResponse();
Stream httpResponseStream = httpResponse.GetResponseStream();
var newStream = new MemoryStream();
httpResponseStream.CopyTo(newStream);
httpResponse.Close();
newStream.Position = 0;
return(newStream);
}
public virtual bool TextToFile(string text, string filePath, SpeechLocaleOptions locale, VoiceName voiceName, GenderOptions voiceType, AudioOutputFormatOptions outputFormat)
{
if (File.Exists(filePath))
{
return(true);
}
//can only send 1000 characters at a time
//speech xml request = 178 characters
var textLimit = 822;
var textParts = (text.Length > textLimit)
? SplitToLength(text, textLimit)
: new List <string> {
text
};
using (var fileStream = new FileStream(filePath, FileMode.Append, FileAccess.Write, FileShare.Write))
{
foreach (var t in textParts)
{
using (var dataStream = TextToSpeech(t, locale, voiceName, voiceType, outputFormat))
{
if (dataStream != null)
{
dataStream.CopyTo(fileStream);
}
}
}
fileStream.Close();
}
return(true);
}
public virtual string GetAudioContent(string text, SpeechLocaleOptions locale, VoiceName voiceName, GenderOptions voiceType)
{
var langValue = locale.MemberAttrValue();
var namespaceAttr = $"xmlns=\"http://www.w3.org/2001/10/synthesis\"";
var versionAttr = $"version=\"1.0\"";
var langAttr = $"xml:lang=\"{langValue}\"";
var genderAttr = $"xmlns:gender=\"{voiceType}\"";
var nameAttr = $"name=\"Microsoft Server Speech Text to Speech Voice {voiceName.MemberAttrValue()}\"";
var speakTag = $"<speak {versionAttr} {namespaceAttr} {langAttr}><voice {nameAttr}>{text}</voice></speak>";
return(speakTag);
}
public virtual Stream TextToSpeech(string text, SpeechLocaleOptions locale, VoiceName voiceName, GenderOptions voiceType, AudioOutputFormatOptions outputFormat)
{
return(PolicyService.ExecuteRetryAndCapture400Errors(
"SpeechService.TextToSpeech",
ApiKeys.SpeechRetryInSeconds,
() =>
{
var result = SpeechRepository.TextToSpeech(text, locale, voiceName, voiceType, outputFormat);
return result;
},
null));
}
public virtual Task <SpeechToTextResponse> SpeechToTextAsync(Stream audioStream, ScenarioOptions scenario,
SpeechLocaleOptions locale, SpeechOsOptions os, Guid fromDeviceId, int maxnbest = 1, int profanitycheck = 1)
{
return(PolicyService.ExecuteRetryAndCapture400Errors(
"SpeechService.SpeechToTextAsync",
ApiKeys.SpeechRetryInSeconds,
() =>
{
var result = SpeechRepository.SpeechToTextAsync(audioStream, scenario, locale, os, fromDeviceId, maxnbest, profanitycheck);
return result;
},
null));
}
public virtual async Task <Stream> TextToSpeechAsync(string text, SpeechLocaleOptions locale, VoiceName voiceName, GenderOptions voiceType, AudioOutputFormatOptions outputFormat)
{
var response = await RepositoryClient.GetAudioStreamAsync(
ApiKeys.SpeechEndpoint,
text,
locale,
voiceName,
voiceType,
outputFormat,
GetSpeechToken());
return(response);
}
public virtual async Task <Stream> GetAudioStreamAsync(string url, string text, SpeechLocaleOptions locale, VoiceName voiceName, GenderOptions voiceType, AudioOutputFormatOptions outputFormat, string token)
{
HttpClientHandler handler = new HttpClientHandler()
{
CookieContainer = new CookieContainer(), UseProxy = false
};
HttpClient client = new HttpClient(handler);
client.DefaultRequestHeaders.Clear();
var Headers = GetAudioHeaders(token, outputFormat);
foreach (var header in Headers)
{
client.DefaultRequestHeaders.TryAddWithoutValidation(header.Key, header.Value);
}
var request = new HttpRequestMessage(HttpMethod.Post, url)
{
Content = new StringContent(GetAudioContent(text, locale, voiceName, voiceType))
};
var response = await client.SendAsync(request, HttpCompletionOption.ResponseHeadersRead, CancellationToken.None);
if (!response.IsSuccessStatusCode)
{
throw new Exception($"({response.StatusCode}) {response.ReasonPhrase}");
}
return(await response.Content.ReadAsStreamAsync());
}
/// <summary>
/// This transcribes voice queries
/// </summary>
/// <param name="audioStream">Audio stream</param>
/// <param name="scenario">The context for performing a recognition.</param>
/// <param name="locale">Language code of the audio content in IETF RFC 5646. Case does not matter. </param>
/// <param name="os">Operating system the client is running on. This is an open field but we encourage clients to use it consistently across devices and applications.</param>
/// <param name="fromDeviceId">A globally unique device identifier of the device making the request.</param>
/// <param name="maxnbest">Maximum number of results the voice application API should return. The default is 1. The maximum is 5. Example: maxnbest=3</param>
/// <param name="profanitycheck">Scan the result text for words included in an offensive word list. If found, the word will be delimited by bad word tag. Example: result.profanity=1 (0 means off, 1 means on, default is 1.)</param>
public virtual async Task <SpeechToTextResponse> SpeechToTextAsync(Stream audioStream, ScenarioOptions scenario, SpeechLocaleOptions locale, SpeechOsOptions os, Guid fromDeviceId, int maxnbest = 1, int profanitycheck = 1)
{
string url = GetSpeechToTextUrl(scenario, locale, os, fromDeviceId, maxnbest, profanitycheck);
string token = GetSpeechToken();
byte[] data = RepositoryClient.GetByteArray(audioStream);
var response = await RepositoryClient.SendAsync(ApiKeys.Speech, url, data, contentType, "POST", token, true, "speech.platform.bing.com");
return(JsonConvert.DeserializeObject <SpeechToTextResponse>(response));
}
protected virtual string GetSpeechToTextUrl(ScenarioOptions scenario, SpeechLocaleOptions locale, SpeechOsOptions os, Guid fromDeviceId, int maxnbest, int profanitycheck)
{
return($"{ApiKeys.SpeechEndpoint}recognize?version=3.0&scenarios={scenario}&appid=D4D52672-91D7-4C74-8AD8-42B1D98141A5&requestid={Guid.NewGuid()}&format=json&locale={locale}&device.os={os}&instanceid={fromDeviceId}&maxnbest={maxnbest}&result.profanitymarkup={profanitycheck}");
}
