add phoneme timestamps to tts.yml (#72)

cartesia-ai · Jan 23, 2025 · e97fecd · e97fecd
1 parent f09bdf2
commit e97fecd
Showing 1 changed file with 19 additions and 1 deletion.
diff --git a/fern/definition/tts.yml b/fern/definition/tts.yml
@@ -124,7 +124,7 @@ channel:
       display-name: "Receive"
       docs: |
         The server will send you back a stream of messages with the same `context_id` as your request.
-        The messages can be of type `chunk`, `timestamps`, `error`, or `done`.
+        The messages can be of type `chunk`, `timestamps`, `phoneme_timestamps` `error`, or `done`.
       origin: server
       body: WebSocketResponse
 
@@ -219,6 +219,7 @@ types:
       done: WebSocketDoneResponse
       timestamps: WebSocketTimestampsResponse
       error: WebSocketErrorResponse
+      phoneme_timestamps: WebSocketPhonemeTimestampsResponse
 
   WebSocketErrorResponse:
     extends: WebSocketBaseResponse
@@ -236,9 +237,15 @@ types:
     properties:
       word_timestamps: optional<WordTimestamps>
 
+  WebSocketPhonemeTimestampsResponse:
+    extends: WebSocketBaseResponse
+    properties:
+      phoneme_timestamps: optional<PhonemeTimestamps>
+
   WebSocketTTSOutput:
     properties:
       word_timestamps: optional<WordTimestamps>
+      phoneme_timestamps: optional<PhonemeTimestamps>
       audio: optional<unknown>
       context_id: optional<ContextID>
 
@@ -252,6 +259,12 @@ types:
       start: list<double>
       end: list<double>
 
+  PhonemeTimestamps:
+    properties:
+      phonemes: list<string>
+      start: list<double>
+      end: list<double>
+
   WebSocketDoneResponse:
     extends: WebSocketBaseResponse
 
@@ -290,6 +303,10 @@ types:
         type: optional<boolean>
         docs: |
           Whether to return word-level timestamps.
+      add_phoneme_timestamps:
+        type: optional<boolean>
+        docs: |
+          Whether to return phoneme-level timestamps.
 
   WebSocketRawOutputFormat:
     properties:
@@ -319,6 +336,7 @@ types:
       duration: optional<integer>
       language: optional<string>
       add_timestamps: optional<boolean>
+      add_phoneme_timestamps: optional<boolean>
       continue: optional<boolean>
       context_id: optional<string>