Add output_type to the RTVI bot-output message

changelog/6351.added.md

@@ -0,0 +1 @@
+- Added `output_type` to the RTVI `bot-output` message, the counterpart of `input_type` on `user-input`. It says what kind of text the message holds: a sentence, a word, a backchannel, or a custom type such as `"code"`.

changelog/6351.deprecated.md

@@ -0,0 +1 @@
+- Deprecated the `text_type` field of the RTVI `bot-output` message. Read `output_type` instead, which carries the same value. The bot sends `output_type`, `text_type` and the deprecated `aggregated_by` until 2.0.0.

docs/architecture/word-trecking/rtvi-integration.md

@@ -146,7 +146,7 @@ onBotOutput: (data) => {
   }
   // Anything else (including spoken_status "new") → render a new bubble element
   this.addConversationMessage(
-    data.text, 'bot', data.text_type, data.segment_id,
+    data.text, 'bot', data.output_type, data.segment_id,
   );
 }
 
@@ -159,7 +159,7 @@ highlightSpokenText(data) {
 }
 ```
 
-`data.text_type` carries the segment type, so the client also renders a `code` segment
+`data.output_type` carries the segment type, so the client also renders a `code` segment
 as a syntax-highlighted `<pre>` block and a `link` segment as an anchor — **without parsing
 any tags itself**.
 

scripts/deprecations/deprecations.json

@@ -1461,8 +1461,19 @@
       "deprecated_in": "1.13.0",
       "removed_in": "2.0.0",
       "relation": "use_existing",
-      "replacement": "text_type",
-      "message": "Use ``text_type`` instead. Will be removed in 2.0.0.",
+      "replacement": "output_type",
+      "message": "Use ``output_type`` instead. Will be removed in 2.0.0.",
+      "location": "pipecat/processors/frameworks/rtvi/models.py"
+    },
+    {
+      "subject": "BotOutputMessageData.text_type",
+      "module": "pipecat.processors.frameworks.rtvi.models",
+      "kind": "parameter",
+      "deprecated_in": "1.14.0",
+      "removed_in": "2.0.0",
+      "relation": "use_existing",
+      "replacement": "output_type",
+      "message": "Use ``output_type`` instead. Will be removed in 2.0.0.",
       "location": "pipecat/processors/frameworks/rtvi/models.py"
     },
     {

src/pipecat/evals/events.py

@@ -610,7 +610,7 @@ def _user_input_event(self, data: dict) -> dict | None:
 
     def _bot_backchannel_event(self, data: dict) -> dict | None:
         """The bot's backchannel, from the ``bot-output`` reporting it was spoken to the end."""
-        if data.get("text_type") != "backchannel" or data.get("spoken_status") != "completed":
+        if data.get("output_type") != "backchannel" or data.get("spoken_status") != "completed":
             return None
         progress = data.get("spoken_progress") or {}
         return {

src/pipecat/processors/frameworks/rtvi/models.py

@@ -505,11 +505,17 @@ class BotOutputMessageData(TextMessageData):
     only sees the fields relevant to its version.
 
     Parameters:
-        text_type: What form the text is in (e.g., sentence, code, etc.).
-        aggregated_by: What form the text is in, always the same as ``text_type``.
+        output_type: What form the text is in (e.g., sentence, code, etc.).
+        text_type: What form the text is in, always the same as ``output_type``.
+
+            .. deprecated:: 1.14.0
+                Use ``output_type`` instead.
+                Will be removed in 2.0.0.
+
+        aggregated_by: What form the text is in, always the same as ``output_type``.
 
             .. deprecated:: 1.13.0
-                Use ``text_type`` instead.
+                Use ``output_type`` instead.
                 Will be removed in 2.0.0.
 
         segment_id: ID of the source AggregatedTextFrame.
@@ -523,7 +529,8 @@ class BotOutputMessageData(TextMessageData):
             Present when ``will_be_spoken`` is ``True``.
     """
 
-    text_type: TextType | str
+    output_type: TextType | str
+    text_type: TextType | str | None = None
     aggregated_by: TextType | str | None = None
     segment_id: int | None = None
     # v1 field (protocol 1.4.x)
@@ -535,13 +542,13 @@ class BotOutputMessageData(TextMessageData):
 
     @model_validator(mode="before")
     @classmethod
-    def _sync_text_type(cls, data: Any) -> Any:
-        """Fill whichever of ``text_type`` and ``aggregated_by`` is missing from the other."""
+    def _sync_output_type(cls, data: Any) -> Any:
+        """Fill whichever of ``output_type`` and its deprecated names are missing."""
         if isinstance(data, dict):
-            if "text_type" not in data and "aggregated_by" in data:
-                data = {**data, "text_type": data["aggregated_by"]}
-            elif "aggregated_by" not in data and "text_type" in data:
-                data = {**data, "aggregated_by": data["text_type"]}
+            names = ("output_type", "text_type", "aggregated_by")
+            given = [data[name] for name in names if name in data]
+            if given:
+                data = {**dict.fromkeys(names, given[0]), **data}
         return data
 
 

src/pipecat/processors/frameworks/rtvi/observer.py

@@ -913,7 +913,7 @@ async def _handle_aggregated_progress(self, frame: AggregatedTextProgressFrame):
                 data=RTVI.BotOutputMessageData(
                     text=text,
                     will_be_spoken=True,
-                    text_type=text_type,
+                    output_type=text_type,
                     segment_id=frame.segment_id,
                     spoken_status=spoken_status,
                     spoken_progress=RTVI.SpokenProgressData(
@@ -1007,7 +1007,7 @@ async def _send_aggregated_llm_text(self, frame: AggregatedTextFrame):
                 text=text,
                 spoken=isTTS,
                 will_be_spoken=will_be_spoken,
-                text_type=text_type,
+                output_type=text_type,
                 segment_id=segment_id,
                 spoken_status=spoken_status,
                 spoken_progress=progress,

tests/test_evals_session.py

@@ -243,12 +243,12 @@ def msg(msg_type, data):
                 message={"label": RTVI.MESSAGE_LABEL, "type": msg_type, "data": data}
             )
 
-        def bot_output(text_type, status, accumulated="Mm-hmm."):
+        def bot_output(output_type, status, accumulated="Mm-hmm."):
             return msg(
                 "bot-output",
                 {
                     "text": "Mm-hmm.",
-                    "text_type": text_type,
+                    "output_type": output_type,
                     "spoken_status": status,
                     "spoken_progress": {"accumulated_text": accumulated, "remaining_text": ""},
                 },

tests/test_rtvi_observer_pushes.py

@@ -434,7 +434,7 @@ async def test_a_backchannel_is_sent_as_output_and_tts_text(self):
             [RTVI.BotOutputMessage, RTVI.BotTTSTextMessage, RTVI.BotOutputMessage],
         )
         self.assertEqual(
-            [(data.text_type, data.spoken_status) for data in self._sent_outputs()],
+            [(data.output_type, data.spoken_status) for data in self._sent_outputs()],
             [(TextType.BACKCHANNEL, "new"), (TextType.BACKCHANNEL, "in-progress")],
         )
         self.assertEqual(messages[1].data.text, "Mm-hmm,")

tests/test_text_type_deprecations.py

@@ -127,11 +127,12 @@ def test_tts_service_transformer_aggregation_type():
     assert tts._text_transforms == []
 
 
-@pytest.mark.parametrize("field", ["text_type", "aggregated_by"])
-def test_bot_output_message_carries_both_fields(field):
+@pytest.mark.parametrize("field", ["output_type", "text_type", "aggregated_by"])
+def test_bot_output_message_carries_every_name(field):
     data = RTVI.BotOutputMessageData(text="Hello.", **{field: "status"})
     assert data.model_dump(exclude_none=True) == {
         "text": "Hello.",
+        "output_type": "status",
         "text_type": "status",
         "aggregated_by": "status",
     }