public static string WrapAmazonPitchParam(this TTSPitch pitch, string text)
{
int pitchShift;
switch (pitch)
{
case TTSPitch.X_Low:
case TTSPitch.Low:
case TTSPitch.High:
case TTSPitch.X_High:
//proceed
break;
case TTSPitch.Medium:
return(text);
case TTSPitch.Unassigned:
goto case TTSPitch.Medium;
default:
BGC.Debug.LogError($"TTS Pitch not supported {pitch}");
goto case TTSPitch.Unassigned;
}
//pitchShift = (int)Math.Round(100 * Math.Pow(2, pitch.GetSemitoneShift() / 12.0)) - 100;
pitchShift = (int)Math.Round(100 * (pitch.GetSemitoneShift() / 12.0));
return($"<prosody pitch=\"{pitchShift:+0;-#}%\">{text}</prosody>");
}
public static double GetSemitoneShift(this TTSPitch pitch)
{
switch (pitch)
{
case TTSPitch.X_Low: return(-10);
case TTSPitch.Low: return(-5);
case TTSPitch.Medium: return(0);
case TTSPitch.High: return(5);
case TTSPitch.X_High: return(10);
case TTSPitch.Unassigned:
goto case TTSPitch.Medium;
default:
BGC.Debug.LogError($"TTS Pitch not supported {pitch}");
goto case TTSPitch.Unassigned;
}
}
public async Task <AudioRequest> TTSRequest(TTSVoice voicePreference, TTSPitch pitchPreference, Effect effectsChain, string[] splitTTSText)
{
if (splitTTSText.Any(x => x.Contains('/') || x.Contains('!')))
{
List <AudioRequest> audioRequestSegments = new List <AudioRequest>();
//Complex parsing
StringBuilder stringbuilder = new StringBuilder();
foreach (string ttsWord in splitTTSText)
{
if (ttsWord.Contains('/') || ttsWord.Contains('!'))
{
foreach (string ttsWordSegment in SplitStringByCommandRegex(ttsWord))
{
if (ttsWordSegment.StartsWith('/'))
{
//Sound Effect
SoundEffect soundEffect = soundEffectSystem.GetSoundEffectByAlias(ttsWordSegment);
if (soundEffect is null)
{
//Unrecognized, append as is
stringbuilder.Append(ttsWordSegment);
}
else
{
//Output current
if (stringbuilder.Length > 0)
{
string filename = await GetSynthSpeech(stringbuilder.ToString(), voicePreference, pitchPreference);
audioRequestSegments.Add(new AudioFileRequest(filename, effectsChain));
stringbuilder.Clear();
}
audioRequestSegments.Add(new SoundEffectRequest(soundEffect));
}
}
else if (ttsWordSegment.StartsWith('!'))
{
//Command
AudioRequest request = AudioRequest.ParseCommand(ttsWordSegment.ToLower());
if (request is null)
{
//Unrecognized, append as is
stringbuilder.Append(ttsWordSegment);
}
else
{
//Output current
if (stringbuilder.Length > 0)
{
string filename = await GetSynthSpeech(stringbuilder.ToString(), voicePreference, pitchPreference);
audioRequestSegments.Add(new AudioFileRequest(filename, effectsChain));
stringbuilder.Clear();
}
audioRequestSegments.Add(request);
}
}
else
{
stringbuilder.Append(ttsWordSegment);
}
}
if (stringbuilder.Length > 0)
{
stringbuilder.Append(' ');
}
}
else
{
stringbuilder.Append(ttsWord);
stringbuilder.Append(' ');
}
}
if (stringbuilder.Length > 0)
{
string filename = await GetSynthSpeech(stringbuilder.ToString(), voicePreference, pitchPreference);
audioRequestSegments.Add(new AudioFileRequest(filename, effectsChain));
stringbuilder.Clear();
}
return(new ConcatenatedAudioRequest(audioRequestSegments));
}
else
{
//Simple parsing
string ttsSpeech = string.Join(' ', splitTTSText);
string filename = await GetSynthSpeech(ttsSpeech, voicePreference, pitchPreference);
return(new AudioFileRequest(filename, effectsChain));
}
}
public Task <AudioRequest> TTSRequest(TTSVoice voicePreference, TTSPitch pitchPreference, Effect effectsChain, string ttsText) =>
TTSRequest(voicePreference, pitchPreference, effectsChain, ttsText.Split(' ', options: StringSplitOptions.RemoveEmptyEntries));
protected async Task <string> GetGoogleSynthSpeech(string text, TTSVoice voicePreference, TTSPitch pitchPreference, string filename = null)
{
VoiceSelectionParams voice = voicePreference.GetGoogleVoiceSelectionParams();
AudioConfig config = new AudioConfig
{
AudioEncoding = AudioEncoding.Mp3,
Pitch = pitchPreference.GetSemitoneShift()
};
//TTS
SynthesisInput input = new SynthesisInput
{
Ssml = PrepareGoogleSSML(text)
};
// Perform the Text-to-Speech request, passing the text input
// with the selected voice parameters and audio file type
GoogleSynthesizeSpeechResponse response = await googleClient.SynthesizeSpeechAsync(input, voice, config);
// Write the binary AudioContent of the response to file.
string filepath;
if (string.IsNullOrWhiteSpace(filename))
{
filepath = Path.Combine(TTSFilesPath, $"{Guid.NewGuid()}.mp3");
}
else
{
filepath = Path.Combine(TTSFilesPath, $"{filename}.mp3");
}
using (Stream file = new FileStream(filepath, FileMode.Create))
{
response.AudioContent.WriteTo(file);
}
return(filepath);
}
protected async Task <string> GetAmazonSynthSpeech(string text, TTSVoice voicePreference, TTSPitch pitchPreference, string filename = null)
{
AmazonSynthesizeSpeechRequest synthesisRequest = voicePreference.GetAmazonTTSSpeechRequest();
synthesisRequest.TextType = TextType.Ssml;
synthesisRequest.Text = PrepareAmazonSSML(text, pitchPreference);
// Perform the Text-to-Speech request, passing the text input
// with the selected voice parameters and audio file type
AmazonSynthesizeSpeechResponse synthesisResponse = await amazonClient.SynthesizeSpeechAsync(synthesisRequest);
// Write the binary AudioContent of the response to file.
string filepath;
if (string.IsNullOrWhiteSpace(filename))
{
filepath = Path.Combine(TTSFilesPath, $"{Guid.NewGuid()}.mp3");
}
else
{
filepath = Path.Combine(TTSFilesPath, $"{filename}.mp3");
}
using (Stream file = new FileStream(filepath, FileMode.Create))
{
await synthesisResponse.AudioStream.CopyToAsync(file);
await file.FlushAsync();
file.Close();
}
return(filepath);
}
protected async Task <string> GetSynthSpeech(string text, TTSVoice voicePreference, TTSPitch pitchPreference)
{
switch (voicePreference.GetTTSService())
{
case TTSService.Amazon:
return(await GetAmazonSynthSpeech(text, voicePreference, pitchPreference));
case TTSService.Google:
return(await GetGoogleSynthSpeech(text, voicePreference, pitchPreference));
default:
communication.SendErrorMessage($"Unsupported TTSVoice for TTSService {voicePreference}");
goto case TTSService.Google;
}
}
