Update to enable override to text to speech

ServiceStack · Oct 14, 2024 · cf900b1 · cf900b1
1 parent c8d6ace
commit cf900b1
Show file tree

Hide file tree

Showing 4 changed files with 17 additions and 4 deletions.
diff --git a/AiServer.ServiceInterface/SpeechServices.cs b/AiServer.ServiceInterface/SpeechServices.cs
@@ -60,14 +60,15 @@ public async Task<object> Any(QueueSpeechToText request)
 
     public async Task<object> Any(TextToSpeech request)
     {
+        var model = request.Model ?? "text-to-speech";
         var diffRequest = new CreateGeneration
         {
             Request = new()
             {
-                Model = "text-to-speech",
+                Model = model,
                 Seed = request.Seed,
                 TaskType = AiTaskType.TextToSpeech,
-                PositivePrompt = request.Text
+                PositivePrompt = request.Input
             }
         };
 

diff --git a/AiServer.ServiceModel/Generations.cs b/AiServer.ServiceModel/Generations.cs
@@ -40,13 +40,17 @@ public class TextToSpeech : IGeneration, IReturn<GenerationResponse>
     [ApiMember(Description = "The text to be converted to speech")]
     [Description("The text to be converted to speech")]
     [Required]
-    public string Text { get; set; }
+    public string Input { get; set; }
 
     [ApiMember(Description = "Optional seed for reproducible results in speech generation")]
     [Description("Optional seed for reproducible results in speech generation")]
     [Range(0, int.MaxValue)]
     public int? Seed { get; set; }
 
+    [ApiMember(Description = "Optional specific model and voice to use for speech generation")]
+    [Description("Optional specific model and voice to use for speech generation")]
+    public string? Model { get; set; }
+
     [ApiMember(Description = "Optional client-provided identifier for the request")]
     [Description("Optional client-provided identifier for the request")]
     public string? RefId { get; set; }

diff --git a/AiServer.Tests/SpeechToTextTests.cs b/AiServer.Tests/SpeechToTextTests.cs
@@ -47,7 +47,7 @@ public async Task Can_generate_speech()
         {
             response = await client.ApiAsync(new TextToSpeech
             {
-                Text = "This is a test of synchronous text to speech generation."
+                Input = "This is a test of synchronous text to speech generation."
             });
             response.ThrowIfError();
         }

diff --git a/AiServer/wwwroot/lib/data/media-models.json b/AiServer/wwwroot/lib/data/media-models.json
@@ -108,6 +108,14 @@
       "ComfyUI": "high:en_US-lessac"
     }
   },
+  {
+    "id": "tts_echo",
+    "name": "TTS Echo",
+    "modelType": "TextToSpeech",
+    "apiModels": {
+      "OpenAI": "tts-1:echo"
+    }
+  },
   {
     "id": "text-to-speech",
     "name": "Text to Speech",