livekit-plugins-cartesia 0.3.0__py3-none-any.whl → 0.4.0__py3-none-any.whl

livekit/plugins/cartesia/models.py

@@ -11,4 +11,4 @@ TTSEncoding = Literal[
 
 TTSModels = Literal["sonic-english", "sonic-multilingual"]
 TTSLanguages = Literal["en", "es", "fr", "de", "pt", "zh", "ja"]
-TTSDefaultVoiceId = "b7d50908-b17c-442d-ad8d-810c63997ed9"
+TTSDefaultVoiceId = "c2ac25f9-ecc4-4f56-9095-651354df60c0"

livekit/plugins/cartesia/tts.py

@@ -19,6 +19,7 @@ import base64
 import json
 import os
 from dataclasses import dataclass
+from typing import Any
 
 import aiohttp
 from livekit.agents import tokenize, tts, utils
@@ -42,7 +43,6 @@ class _TTSOptions:
     voice: str | list[float]
     api_key: str
     language: str
-    word_tokenizer: tokenize.WordTokenizer
 
 
 class TTS(tts.TTS):
@@ -56,9 +56,6 @@ class TTS(tts.TTS):
         sample_rate: int = 24000,
         api_key: str | None = None,
         http_session: aiohttp.ClientSession | None = None,
-        word_tokenizer: tokenize.WordTokenizer = tokenize.basic.WordTokenizer(
-            ignore_punctuation=False
-        ),
     ) -> None:
         super().__init__(
             capabilities=tts.TTSCapabilities(streaming=True),
@@ -77,7 +74,6 @@ class TTS(tts.TTS):
             sample_rate=sample_rate,
             voice=voice,
             api_key=api_key,
-            word_tokenizer=word_tokenizer,
         )
         self._session = http_session
 
@@ -145,7 +141,9 @@ class SynthesizeStream(tts.SynthesizeStream):
     ):
         super().__init__()
         self._opts, self._session = opts, session
-        self._buf = ""
+        self._sent_tokenizer_stream = tokenize.basic.SentenceTokenizer(
+            min_sentence_len=BUFFERED_WORDS_COUNT
+        ).stream()
 
     @utils.log_exceptions(logger=logger)
     async def _main_task(self) -> None:
@@ -176,51 +174,29 @@ class SynthesizeStream(tts.SynthesizeStream):
 
     async def _run_ws(self, ws: aiohttp.ClientWebSocketResponse) -> None:
         request_id = utils.shortuuid()
-        pending_segments = []
 
-        async def send_task():
+        async def sentence_stream_task():
             base_pkt = _to_cartesia_options(self._opts)
-
-            def _new_segment():
-                segment_id = utils.shortuuid()
-                pending_segments.append(segment_id)
-                return segment_id
-
-            current_segment_id: str | None = _new_segment()
-
+            async for ev in self._sent_tokenizer_stream:
+                token_pkt = base_pkt.copy()
+                token_pkt["context_id"] = request_id
+                token_pkt["transcript"] = ev.token + " "
+                token_pkt["continue"] = True
+                await ws.send_str(json.dumps(token_pkt))
+
+            end_pkt = base_pkt.copy()
+            end_pkt["context_id"] = request_id
+            end_pkt["transcript"] = " "
+            end_pkt["continue"] = False
+            await ws.send_str(json.dumps(end_pkt))
+
+        async def input_task():
             async for data in self._input_ch:
                 if isinstance(data, self._FlushSentinel):
-                    if current_segment_id is None:
-                        continue
-
-                    end_pkt = base_pkt.copy()
-                    end_pkt["context_id"] = current_segment_id
-                    end_pkt["transcript"] = self._buf + " "
-                    end_pkt["continue"] = False
-                    await ws.send_str(json.dumps(end_pkt))
-
-                    current_segment_id = None
-                    self._buf = ""
-                elif data:
-                    if current_segment_id is None:
-                        current_segment_id = _new_segment()
-
-                    self._buf += data
-                    words = self._opts.word_tokenizer.tokenize(text=self._buf)
-                    if len(words) < BUFFERED_WORDS_COUNT + 1:
-                        continue
-
-                    data = self._opts.word_tokenizer.format_words(words[:-1]) + " "
-                    self._buf = words[-1]
-
-                    token_pkt = base_pkt.copy()
-                    token_pkt["context_id"] = current_segment_id
-                    token_pkt["transcript"] = data
-                    token_pkt["continue"] = True
-                    await ws.send_str(json.dumps(token_pkt))
-
-            if len(pending_segments) == 0:
-                await ws.close()
+                    self._sent_tokenizer_stream.flush()
+                    continue
+                self._sent_tokenizer_stream.push_text(data)
+            self._sent_tokenizer_stream.end_input()
 
         async def recv_task():
             audio_bstream = utils.audio.AudioByteStream(
@@ -243,6 +219,7 @@ class SynthesizeStream(tts.SynthesizeStream):
 
                 data = json.loads(msg.data)
                 segment_id = data.get("context_id")
+                # Once we receive audio for a segment, we can start a new segment
                 if data.get("data"):
                     b64data = base64.b64decode(data["data"])
                     for frame in audio_bstream.write(b64data):
@@ -263,8 +240,7 @@ class SynthesizeStream(tts.SynthesizeStream):
                             )
                         )
 
-                    pending_segments.remove(segment_id)
-                    if len(pending_segments) == 0 and self._input_ch.closed:
+                    if segment_id == request_id:
                         # we're not going to receive more frames, close the connection
                         await ws.close()
                         break
@@ -272,7 +248,8 @@ class SynthesizeStream(tts.SynthesizeStream):
                     logger.error("unexpected Cartesia message %s", data)
 
         tasks = [
-            asyncio.create_task(send_task()),
+            asyncio.create_task(input_task()),
+            asyncio.create_task(sentence_stream_task()),
             asyncio.create_task(recv_task()),
         ]
 
@@ -282,8 +259,8 @@ class SynthesizeStream(tts.SynthesizeStream):
             await utils.aio.gracefully_cancel(*tasks)
 
 
-def _to_cartesia_options(opts: _TTSOptions) -> dict:
-    voice: dict = {}
+def _to_cartesia_options(opts: _TTSOptions) -> dict[str, Any]:
+    voice: dict[str, Any] = {}
     if isinstance(opts.voice, str):
         voice["mode"] = "id"
         voice["id"] = opts.voice

livekit/plugins/cartesia/version.py

@@ -12,4 +12,4 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-__version__ = "0.3.0"
+__version__ = "0.4.0"

{livekit_plugins_cartesia-0.3.0.dist-info → livekit_plugins_cartesia-0.4.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: livekit-plugins-cartesia
-Version: 0.3.0
+Version: 0.4.0
 Summary: LiveKit Agents Plugin for Cartesia
 Home-page: https://github.com/livekit/agents
 License: Apache-2.0

livekit_plugins_cartesia-0.4.0.dist-info/RECORD

@@ -0,0 +1,10 @@
+livekit/plugins/cartesia/__init__.py,sha256=BUfWY_evL5dUHn9hBDQVor6ssctDKQfbQfZy5SWndN8,926
+livekit/plugins/cartesia/log.py,sha256=4Mnhjng_DU1dIWP9IWjIQGZ67EV3LnQhWMWCHVudJbo,71
+livekit/plugins/cartesia/models.py,sha256=ZoSyV2ap_LqAIgvBvkmukkPxQR9DfKb3Z3oHtWxMiVg,335
+livekit/plugins/cartesia/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+livekit/plugins/cartesia/tts.py,sha256=sdiiWinOZR5EBkQFwa3GZAGrkgzXY1-aSRiDZ34K8ww,9527
+livekit/plugins/cartesia/version.py,sha256=yelanl1wEXtgUH0CzoNVXfi2yTc2hElSzuAhULFzANc,600
+livekit_plugins_cartesia-0.4.0.dist-info/METADATA,sha256=BGgicrqKsylOpTbUcRG0B4DZF2qnaERI9q7qwIRLN7s,1252
+livekit_plugins_cartesia-0.4.0.dist-info/WHEEL,sha256=HiCZjzuy6Dw0hdX5R3LCFPDmFS4BWl8H-8W39XfmgX4,91
+livekit_plugins_cartesia-0.4.0.dist-info/top_level.txt,sha256=OoDok3xUmXbZRvOrfvvXB-Juu4DX79dlq188E19YHoo,8
+livekit_plugins_cartesia-0.4.0.dist-info/RECORD,,

{livekit_plugins_cartesia-0.3.0.dist-info → livekit_plugins_cartesia-0.4.0.dist-info}/WHEEL

@@ -1,5 +1,5 @@
 Wheel-Version: 1.0
-Generator: setuptools (72.1.0)
+Generator: setuptools (72.2.0)
 Root-Is-Purelib: true
 Tag: py3-none-any
 

livekit_plugins_cartesia-0.3.0.dist-info/RECORD

@@ -1,10 +0,0 @@
-livekit/plugins/cartesia/__init__.py,sha256=BUfWY_evL5dUHn9hBDQVor6ssctDKQfbQfZy5SWndN8,926
-livekit/plugins/cartesia/log.py,sha256=4Mnhjng_DU1dIWP9IWjIQGZ67EV3LnQhWMWCHVudJbo,71
-livekit/plugins/cartesia/models.py,sha256=T1iPQ18h4-o5rgSW236PDc73qp5zR9k4r_qNCl3XPWc,335
-livekit/plugins/cartesia/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-livekit/plugins/cartesia/tts.py,sha256=uklD9fIYL8QWUSiyypFDgflkie9VhTu1C-x4YwJcDCU,10283
-livekit/plugins/cartesia/version.py,sha256=G5iYozum4q7UpHwW43F7QfhzUfwcncPxBZ0gmUGsd5I,600
-livekit_plugins_cartesia-0.3.0.dist-info/METADATA,sha256=iJcOyrkQ-0yPK_lYtR-eEbIDav84xlN7DUvwncx7OpQ,1252
-livekit_plugins_cartesia-0.3.0.dist-info/WHEEL,sha256=R0nc6qTxuoLk7ShA2_Y-UWkN8ZdfDBG2B6Eqpz2WXbs,91
-livekit_plugins_cartesia-0.3.0.dist-info/top_level.txt,sha256=OoDok3xUmXbZRvOrfvvXB-Juu4DX79dlq188E19YHoo,8
-livekit_plugins_cartesia-0.3.0.dist-info/RECORD,,