add yandex tts

src/Translumo.Processing/TranslationProcessingService.cs

@@ -52,10 +52,10 @@ public class TranslationProcessingService : IProcessingService, IDisposable
         private long _lastTranslatedTextTicks;
 
         private const float MIN_SCORE_THRESHOLD = 2.1f;
-        
+
         public TranslationProcessingService(ICapturerFactory capturerFactory, IChatTextMediator chatTextMediator, OcrEnginesFactory ocrEnginesFactory,
             TranslatorFactory translationFactory, TtsFactory ttsFactory, TtsConfiguration ttsConfiguration,
-            TextDetectionProvider textProvider, TranslationConfiguration translationConfiguration, OcrGeneralConfiguration ocrConfiguration, 
+            TextDetectionProvider textProvider, TranslationConfiguration translationConfiguration, OcrGeneralConfiguration ocrConfiguration,
             TextResultCacheService textResultCacheService, TextProcessingConfiguration textConfiguration, ILogger<TranslationProcessingService> logger)
         {
             _logger = logger;
@@ -236,7 +236,7 @@ void CapturerEnsureInitialized()
                             continue;
                         }
 
-                        if (_textResultCacheService.IsCached(bestDetected.Text, bestDetected.ValidityScore, sequentialText, 
+                        if (_textResultCacheService.IsCached(bestDetected.Text, bestDetected.ValidityScore, sequentialText,
                                 bestDetected.Language.Asian, out iterationId))
                         {
                             sequentialText = false;
@@ -257,7 +257,7 @@ void CapturerEnsureInitialized()
                     }
 
                     _logger.LogError(ex, $"Screen capture failed (code: {ex.ErrorCode})");
-                    
+
                     _capturer.Dispose();
                     _capturer = null;
                     CapturerEnsureInitialized();
@@ -407,9 +407,15 @@ private void TtsConfigurationOnPropertyChanged(object sender, PropertyChangedEve
             if (e.PropertyName == nameof(_ttsConfiguration.TtsLanguage)
                 || e.PropertyName == nameof(_ttsConfiguration.TtsSystem))
             {
-                _ttsEngine.Dispose();
+                _ttsEngine?.Dispose();
+                _ttsEngine = null;
                 _ttsEngine = _ttsFactory.CreateTtsEngine(_ttsConfiguration);
             }
+            else if (e.PropertyName == nameof(_ttsConfiguration.CurrentVoice)
+                && _ttsEngine != null && _ttsConfiguration.CurrentVoice != null)
+            {
+                _ttsEngine.SetVoice(_ttsConfiguration.CurrentVoice);
+            }
         }
 
         private void OcrGeneralConfigurationOnPropertyChanged(object sender, PropertyChangedEventArgs e)

src/Translumo.TTS/Engines/ITTSEngine.cs

@@ -1,6 +1,10 @@
 namespace Translumo.TTS.Engines;
 
-public interface ITTSEngine: IDisposable
+public interface ITTSEngine : IDisposable
 {
     void SpeechText(string text);
+
+    string[] GetVoices();
+
+    void SetVoice(string voice);
 }

src/Translumo.TTS/Engines/NoneTTSEngine.cs

@@ -10,6 +10,12 @@ public void Dispose()
     {
     }
 
+    public string[] GetVoices() => new[] { "None" };
+
+    public void SetVoice(string voice)
+    {
+    }
+
     public void SpeechText(string text)
     {
     }

src/Translumo.TTS/Engines/SileroTTSEngine.cs

@@ -8,6 +8,7 @@ public class SileroTTSEngine : ITTSEngine
 {
     private dynamic _ipython;
     private dynamic _model;
+    private string[] _voices;
     private string _voice;
     private readonly string _modelPath;
     private readonly PythonEngineWrapper _pythonEngine;
@@ -51,7 +52,8 @@ private void Init()
             _pyObjects.Add(_ipython);
         });
 
-        _voice = ((string[])_model.speakers).First();
+        _voices = (string[])_model.speakers;
+        _voice = _voices.First();
     }
 
     public void SpeechText(string text)
@@ -146,5 +148,9 @@ private string GetModelFullPath(string langCode)
             _ => null
         };
 
+    public string[] GetVoices() => _voices;
+
+    public void SetVoice(string voice) => _voice = _voices.First(x => x.Equals(voice, StringComparison.OrdinalIgnoreCase));
+
     private sealed record ModelDescription(string FileUrl, string WarmUpText);
 }

src/Translumo.TTS/Engines/WindowsTTSEngine.cs

@@ -1,20 +1,25 @@
-using System.Globalization;
+using System.Collections;
+using System.Collections.ObjectModel;
+using System.Globalization;
+using System.Reflection;
 using System.Speech.Synthesis;
 
 namespace Translumo.TTS.Engines;
 
 public class WindowsTTSEngine : ITTSEngine
 {
-    private readonly VoiceInfo _voiceInfo;
+    private VoiceInfo _voiceInfo;
     private readonly SpeechSynthesizer _synthesizer;
+    private readonly ReadOnlyDictionary<string, VoiceInfo> _voices;
 
     public WindowsTTSEngine(string languageCode)
     {
         _synthesizer = new SpeechSynthesizer();
         _synthesizer.SetOutputToDefaultAudioDevice();
         _synthesizer.Rate = 1;
-
-        _voiceInfo = _synthesizer.GetInstalledVoices(new CultureInfo(languageCode)).FirstOrDefault()?.VoiceInfo;
+        SpeechApiReflectionHelper.InjectOneCoreVoices(_synthesizer);
+        _voices = _synthesizer.GetInstalledVoices(new CultureInfo(languageCode)).ToDictionary(x => x.VoiceInfo.Name, x => x.VoiceInfo).AsReadOnly();
+        _voiceInfo = _voices.First().Value;
     }
 
     public void SpeechText(string text)
@@ -36,4 +41,88 @@ public void Dispose()
     {
         _synthesizer.Dispose();
     }
+
+    public string[] GetVoices() => _voices.Keys.ToArray();
+
+    public void SetVoice(string voice) => _voiceInfo = _voices.First(x => x.Key.Equals(voice, StringComparison.OrdinalIgnoreCase)).Value;
+
+    // by default SpeechSynthesizer show not all available voices
+    // https://stackoverflow.com/a/71198211
+    private static class SpeechApiReflectionHelper
+    {
+        private const string PROP_VOICE_SYNTHESIZER = "VoiceSynthesizer";
+        private const string FIELD_INSTALLED_VOICES = "_installedVoices";
+
+        private const string ONE_CORE_VOICES_REGISTRY = @"HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Speech_OneCore\Voices";
+
+        private static readonly Type _objectTokenCategoryType = typeof(SpeechSynthesizer).Assembly
+            .GetType("System.Speech.Internal.ObjectTokens.ObjectTokenCategory")!;
+
+        private static readonly Type _voiceInfoType = typeof(SpeechSynthesizer).Assembly
+            .GetType("System.Speech.Synthesis.VoiceInfo")!;
+
+        private static readonly Type _installedVoiceType = typeof(SpeechSynthesizer).Assembly
+            .GetType("System.Speech.Synthesis.InstalledVoice")!;
+
+
+        public static void InjectOneCoreVoices(SpeechSynthesizer synthesizer)
+        {
+            var voiceSynthesizer = GetProperty(synthesizer, PROP_VOICE_SYNTHESIZER);
+            if (voiceSynthesizer == null)
+                throw new NotSupportedException($"Property not found: {PROP_VOICE_SYNTHESIZER}");
+
+            var installedVoices = GetField(voiceSynthesizer, FIELD_INSTALLED_VOICES) as IList;
+            if (installedVoices == null)
+                throw new NotSupportedException($"Field not found or null: {FIELD_INSTALLED_VOICES}");
+
+            if (_objectTokenCategoryType
+                    .GetMethod("Create", BindingFlags.Static | BindingFlags.NonPublic)?
+                    .Invoke(null, new object?[] { ONE_CORE_VOICES_REGISTRY }) is not IDisposable otc)
+                throw new NotSupportedException($"Failed to call Create on {_objectTokenCategoryType} instance");
+
+            using (otc)
+            {
+                if (_objectTokenCategoryType
+                        .GetMethod("FindMatchingTokens", BindingFlags.Instance | BindingFlags.NonPublic)?
+                        .Invoke(otc, new object?[] { null, null }) is not IList tokens)
+                    throw new NotSupportedException($"Failed to list matching tokens");
+
+                foreach (var token in tokens)
+                {
+                    if (token == null || GetProperty(token, "Attributes") == null)
+                        continue;
+
+                    var voiceInfo =
+                        typeof(SpeechSynthesizer).Assembly
+                            .CreateInstance(_voiceInfoType.FullName!, true,
+                                BindingFlags.Instance | BindingFlags.NonPublic, null,
+                                new object[] { token }, null, null);
+
+                    if (voiceInfo == null)
+                        throw new NotSupportedException($"Failed to instantiate {_voiceInfoType}");
+
+                    var installedVoice =
+                        typeof(SpeechSynthesizer).Assembly
+                            .CreateInstance(_installedVoiceType.FullName!, true,
+                                BindingFlags.Instance | BindingFlags.NonPublic, null,
+                                new object[] { voiceSynthesizer, voiceInfo }, null, null);
+
+                    if (installedVoice == null)
+                        throw new NotSupportedException($"Failed to instantiate {_installedVoiceType}");
+
+                    installedVoices.Add(installedVoice);
+                }
+            }
+        }
+
+        private static object? GetProperty(object target, string propName)
+        {
+            return target.GetType().GetProperty(propName, BindingFlags.Instance | BindingFlags.NonPublic)?.GetValue(target);
+        }
+
+        private static object? GetField(object target, string propName)
+        {
+            return target.GetType().GetField(propName, BindingFlags.Instance | BindingFlags.NonPublic)?.GetValue(target);
+        }
+    }
 }

src/Translumo.TTS/Engines/YandexTTSEngine.cs

@@ -0,0 +1,96 @@
+using NAudio.Wave;
+
+namespace Translumo.TTS.Engines;
+
+public class YandexTTSEngine : ITTSEngine
+{
+    private readonly string _langCode;
+    private string _voice;
+    private CancellationTokenSource _tokenSource;
+    private readonly DirectSoundOut _naudioPlayer;
+    private readonly string _emotion = "neutral"; // neutral|good|evil
+    private const float _speed = 1.0f;
+    private const string _format = "mp3"; // ogg | mp3 | wav
+    private const string _key = "1"; // yandex doesnt validate key, its can have any not empty value
+
+    public YandexTTSEngine(string langCode)
+    {
+        _langCode = langCode;
+        _voice = GetVoices().First();
+        _naudioPlayer = new DirectSoundOut();
+        _tokenSource = new CancellationTokenSource();
+    }
+
+    public void SpeechText(string text)
+    {
+        _tokenSource.Cancel();
+        _tokenSource = new CancellationTokenSource();
+        var currentToken = _tokenSource.Token;
+        Task.Run(() => SpeechTextInternalAsync(text, currentToken));
+    }
+
+    private async Task SpeechTextInternalAsync(string text, CancellationToken token)
+    {
+        using var stream = await RequestAudioAsync(text, token);
+        using var audioFileReader = new Mp3FileReader(stream);
+
+        _naudioPlayer.Stop();
+        _naudioPlayer.Init(audioFileReader);
+        _naudioPlayer.Play();
+
+        var duration = Convert.ToInt32(Math.Round(audioFileReader.TotalTime.TotalMilliseconds, MidpointRounding.ToPositiveInfinity));
+        Task.Delay(duration).Wait(token);
+    }
+
+    private async Task<Stream> RequestAudioAsync(string text, CancellationToken token)
+    {
+        var httpClient = new HttpClient();
+        var httpResponse = await httpClient.GetAsync(BuildUrl(text), token).ConfigureAwait(false);
+        if (httpResponse.IsSuccessStatusCode)
+        {
+            return await httpResponse.Content.ReadAsStreamAsync(token).ConfigureAwait(false);
+        }
+
+        var error = await httpResponse.Content.ReadAsStringAsync(token);
+        throw new InvalidOperationException($"Failed to get sound: '{httpResponse.StatusCode}, {error}'");
+    }
+
+    private string BuildUrl(string text)
+    {
+        var encodedText = Uri.EscapeDataString(text);
+        return $"https://tts.voicetech.yandex.net/generate?text={encodedText}&lang={_langCode}&key={_key}&speaker={_voice}&format={_format}&speed={_speed:#.##}&emotion={_emotion}&quality=hi";
+    }
+
+    public void SetVoice(string voice) =>
+    _voice = GetVoices().FirstOrDefault(x => x.Equals(voice, StringComparison.OrdinalIgnoreCase))
+        ?? throw new InvalidDataException($"Voice '{voice}' doesnt support '{_langCode}' language ");
+
+    public string[] GetVoices() =>
+        SupportedVoices().GetValueOrDefault(_langCode)
+        ?? throw new InvalidDataException($"{nameof(YandexTTSEngine)} doesnt support '{_langCode}' language");
+
+    public void Dispose()
+    {
+        _tokenSource.Cancel();
+        _naudioPlayer.Stop();
+        _naudioPlayer.Dispose();
+    }
+
+    private static Dictionary<string, string[]> SupportedVoices()
+    {
+        var listOfVoices = new[] {
+            // dictor dialog voices: zahar, ermil, filipp,|  jane, omazh, alena
+            "zahar", "ermil",  "dude", "jane", "omazh", "oksana",
+            // additional voices
+            "alyss", "erkanyavas", "ermilov", "kolya", "kostya", "levitan", "nastya", "nick", "sasha", "silaerkan", "smoky", "tanya", "voicesearch", "zhenya", "zombie"
+        };
+
+        return new Dictionary<string, string[]>()
+        {
+            // dictor dialog voices: zahar, ermil, filipp,|  jane, omazh, alena
+            { "ru-RU", listOfVoices},
+            { "en-US", listOfVoices},
+            { "tr-TR", listOfVoices},
+        };
+    }
+}

src/Translumo.TTS/IObserverAvailableVoices.cs

@@ -0,0 +1,6 @@
+namespace Translumo.TTS;
+
+public interface IObserverAvailableVoices
+{
+    void UpdateVoice(IList<string> currentVoices);
+}

src/Translumo.TTS/TTSEngines.cs

@@ -5,5 +5,6 @@ public enum TTSEngines
         None = 0,
         WindowsTTS = 1,
         SileroTTS = 2,
+        YandexTTS = 3,
     }
 }

src/Translumo.TTS/Translumo.TTS.csproj

@@ -7,6 +7,7 @@
 
   <ItemGroup>
     <PackageReference Include="Microsoft.Extensions.Logging.Abstractions" Version="7.0.0" />
+    <PackageReference Include="NAudio" Version="2.2.1" />
     <PackageReference Include="System.Speech" Version="7.0.0" />
   </ItemGroup>
 

src/Translumo.TTS/TtsConfiguration.cs

@@ -1,6 +1,6 @@
-using Translumo.Infrastructure.Language;
+using System.Collections.ObjectModel;
+using Translumo.Infrastructure.Language;
 using Translumo.Utils;
-using Windows.Security.EnterpriseData;
 
 namespace Translumo.TTS;
 
@@ -12,12 +12,14 @@ public class TtsConfiguration : BindableBase
         {
             TtsLanguage = Languages.English,
             TtsSystem = TTSEngines.None,
-            InstalledWinTtsLanguages = new List<Languages>()
+            InstalledWinTtsLanguages = new List<Languages>(),
+            _currentVoice = string.Empty,
         };
 
     private TTSEngines _ttsSystem;
     private Languages _ttsLanguage;
     private List<Languages> _installedWinTtsLanguages;
+    private string _currentVoice;
 
     public TTSEngines TtsSystem
     {
@@ -37,6 +39,12 @@ public Languages TtsLanguage
         }
     }
 
+    public string CurrentVoice
+    {
+        get => _currentVoice;
+        set => SetProperty(ref _currentVoice, value);
+    }
+
     public List<Languages> InstalledWinTtsLanguages
     {
         get => _installedWinTtsLanguages;