private byte[] Synthesis(EdgeTTSSettings settings, string text)
{
var ws = ObtainConnection();
if (ws == null)
{
// Cancelled
return(null);
}
lock (ws)
{
// Send request
var requestId = Guid.NewGuid().ToString().Replace("-", "");
var timestamp = DateTime.UtcNow.ToString("yyyy-MM-ddTHH:mm:ss.fffK");
try
{
ws.SendText(
"Path:speech.config\r\n" +
$"X-RequestId:{requestId}\r\n" +
$"X-Timestamp:{timestamp}\r\n" +
"Content-Type:application/json\r\n" +
"\r\n" +
"{\"context\":{\"synthesis\":{\"audio\":{\"metadataoptions\":{\"sentenceBoundaryEnabled\":\"false\",\"wordBoundaryEnabled\":\"false\"},\"outputFormat\":\"audio-24khz-48kbitrate-mono-mp3\"}}}}\r\n",
_wsCancellationSource
);
ws.SendText(
"Path:ssml\r\n" +
$"X-RequestId:{requestId}\r\n" +
$"X-Timestamp:{timestamp}\r\n" +
"Content-Type:application/ssml+xml\r\n" +
"\r\n" +
"<speak xmlns=\"http://www.w3.org/2001/10/synthesis\" xmlns:mstts=\"http://www.w3.org/2001/mstts\" xmlns:emo=\"http://www.w3.org/2009/10/emotionml\" version=\"1.0\" xml:lang=\"en-US\">" +
$"<voice name=\"{settings.Voice}\">" +
$"<prosody rate=\"{settings.Speed - 100}%\" pitch=\"{(settings.Pitch - 100) / 2}%\" volume=\"{settings.Volume.Clamp(1, 100)}\">" +
text +
"</prosody></voice></speak>\r\n",
_wsCancellationSource
);
}
catch (Exception)
{
ws.Abort();
ws.Dispose();
throw;
}
// Start receiving
var buffer = new MemoryStream();
var session = new WebSocketHelper.Session(ws);
var state = ProtocolState.NotStarted;
while (true)
{
var message = WebSocketHelper.ReceiveNextMessage(session, _wsCancellationSource);
Logger.Debug($"Received WS message\n{message}");
if (message.Type == WebSocketMessageType.Text)
{
if (message.MessageStr.Contains(requestId))
{
switch (state)
{
case ProtocolState.NotStarted:
if (message.MessageStr.Contains("Path:turn.start"))
{
state = ProtocolState.TurnStarted;
}
break;
case ProtocolState.TurnStarted:
if (message.MessageStr.Contains("Path:turn.end"))
{
throw new IOException("Unexpected turn.end");
}
else if (message.MessageStr.Contains("Path:turn.start"))
{
throw new IOException("Turn already started");
}
break;
case ProtocolState.Streaming:
if (message.MessageStr.Contains("Path:turn.end"))
{
// All done
return(buffer.ToArray());
}
else
{
throw new IOException($"Unexpected message during streaming: {message.MessageStr}");
}
default:
throw new ArgumentOutOfRangeException();
}
}
else
{
if (state != ProtocolState.NotStarted)
{
throw new IOException("Unexpected request id during streaming");
}
else
{
// Ignore
}
}
}
else if (message.Type == WebSocketMessageType.Binary)
{
switch (state)
{
case ProtocolState.NotStarted:
// Do nothing
break;
case ProtocolState.TurnStarted:
case ProtocolState.Streaming:
// Parsing message
// The first 2 bytes are the header length
if (message.MessageBinary.Length < 2)
{
throw new IOException("Message too short");
}
var headerLen = (message.MessageBinary[0] << 8) + message.MessageBinary[1];
if (message.MessageBinary.Length < 2 + headerLen)
{
throw new IOException("Message too short");
}
var header = Encoding.UTF8.GetString(message.MessageBinary, 2, headerLen);
if (header.EndsWith("Path:audio\r\n"))
{
if (!header.Contains(requestId))
{
throw new IOException("Unexpected request id during streaming");
}
state = ProtocolState.Streaming;
buffer.Write(message.MessageBinary, 2 + headerLen, message.MessageBinary.Length - 2 - headerLen);
}
else
{
Logger.Warn($"Unexpected message with header {header}");
}
break;
default:
throw new ArgumentOutOfRangeException();
}
}
else if (message.Type == WebSocketMessageType.Close)
{
throw new IOException("Unexpected closing of connection");
}
else
{
throw new ArgumentOutOfRangeException();
}
}
}
}
private byte[] Synthesis(XfyunTTSSettings settings, string text)
{
var apiSecret = settings.ApiSecret;
var apiKey = settings.ApiKey;
var appId = settings.AppId;
if (string.IsNullOrEmpty(apiKey) || string.IsNullOrEmpty(apiSecret) || string.IsNullOrEmpty(appId))
{
Logger.Error(strings.msgErrorEmptyApiSecretKey);
_settingsControl.NotifyEmptyApiKey();
return(null);
}
var base64Text = Convert.ToBase64String(Encoding.UTF8.GetBytes(text));
if (base64Text.Length > 8000)
{
Logger.Error("Convert string too long. No more than 2000 chinese characters.");
return(null);
}
using (var ws = SystemClientWebSocket.CreateClientWebSocket())
{
// Connect
var date = DateTime.UtcNow.ToString("R");
var sign = $"host: tts-api.xfyun.cn\ndate: {date}\nGET /v2/tts HTTP/1.1";
string sha;
using (var hash = new HMACSHA256(Encoding.UTF8.GetBytes(apiSecret)))
{
sha = Convert.ToBase64String(hash.ComputeHash(Encoding.UTF8.GetBytes(sign)));
hash.Clear();
}
var authorization =
$"api_key=\"{apiKey}\"," +
$" algorithm=\"hmac-sha256\"," +
$" headers=\"host date request-line\"," +
$" signature=\"{sha}\"";
var url =
$"wss://tts-api.xfyun.cn/v2/tts?host=tts-api.xfyun.cn" +
$"&date={WebUtility.UrlEncode(date).Replace("+", "%20")}" +
$"&authorization={Convert.ToBase64String(Encoding.UTF8.GetBytes(authorization))}";
try
{
ws.ConnectAsync(new Uri(url), _wsCancellationSource.Token).Wait();
}
catch (AggregateException e)
{
var inner = e.InnerExceptions.First().GetBaseException();
if (inner is WebException webException)
{
var resp = (HttpWebResponse)webException.Response;
string body = null;
using (var stream = resp.GetResponseStream())
{
if (stream != null)
{
var reader = new StreamReader(stream, string.IsNullOrEmpty(resp.CharacterSet) ? Encoding.UTF8 : Encoding.GetEncoding(resp.CharacterSet));
body = reader.ReadToEnd();
}
}
Logger.Error($"Unable to connect to server: {body}");
switch (resp.StatusCode)
{
case HttpStatusCode.Unauthorized:
case HttpStatusCode.Forbidden:
Logger.Error(strings.msgErrorXfyunAuthFail);
return(null);
}
}
throw;
}
// Send request
var request = new TTSRequest
{
Common = new TTSRequest.CommonParam
{
AppId = settings.AppId,
},
Business = new TTSRequest.BusinessParam
{
Voice = settings.Voice,
Pitch = settings.Pitch * 10,
Speed = settings.Speed * 10,
Volume = settings.Volume * 10,
},
Data = new TTSRequest.DataParam
{
Status = 2,
Text = base64Text,
}
};
ws.SendText(JsonConvert.SerializeObject(request), _wsCancellationSource);
// Start receiving
var buffer = new MemoryStream();
var session = new WebSocketHelper.Session(ws);
while (true)
{
var message = WebSocketHelper.ReceiveNextMessage(session, _wsCancellationSource);
Logger.Debug($"Received WS message\n{message}");
if (message.Type == WebSocketMessageType.Text)
{
var resp = JsonConvert.DeserializeObject <TTSResponse>(message.MessageStr);
if (resp.Code == 0)
{
// Success!
if (resp.Data != null)
{
var data = Convert.FromBase64String(resp.Data.Audio);
buffer.Write(data, 0, data.Length);
if (resp.Data.Status == 2)
{
// Complete!
return(buffer.ToArray());
}
}
}
else
{
Logger.Error($"Unexpected response code received: {resp.Code}: {resp.Message}");
switch (resp.Code)
{
case 10005:
case 10313:
Logger.Error(strings.msgErrorXfyunWrongAppId);
break;
case 11200:
case 11201:
Logger.Error(strings.msgErrorXfyunInsufficientApiQuota);
break;
}
return(null);
}
}
else if (message.Type == WebSocketMessageType.Binary)
{
throw new IOException("Unexpected binary message received");
}
else if (message.Type == WebSocketMessageType.Close)
{
throw new IOException("Unexpected closing of connection");
}
else
{
throw new ArgumentOutOfRangeException();
}
}
}
}
