cartesia 1.0.9__tar.gz → 1.0.11__tar.gz

{cartesia-1.0.9 → cartesia-1.0.11}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: cartesia
-Version: 1.0.9
+Version: 1.0.11
 Summary: The official Python library for the Cartesia API.
 Home-page: 
 Author: Cartesia, Inc.

{cartesia-1.0.9 → cartesia-1.0.11}/cartesia/_types.py

@@ -1,4 +1,4 @@
-from typing import List, TypedDict
+from typing import List, TypedDict, Union
 
 from cartesia.utils.deprecated import deprecated
 
@@ -86,7 +86,7 @@ class VoiceControls(TypedDict):
         This is an experimental class and is subject to rapid change in future versions.
     """
 
-    speed: str = ""
+    speed: Union[str, float] = ""
     emotion: List[str] = []
 
 

{cartesia-1.0.9 → cartesia-1.0.11}/cartesia/client.py

@@ -31,6 +31,9 @@ try:
 except ImportError:
     IS_WEBSOCKET_SYNC_AVAILABLE = False
 
+from iterators import TimeoutIterator
+from websockets.sync.client import connect
+
 from cartesia._types import (
     DeprecatedOutputFormatMapping,
     EventType,
@@ -40,8 +43,6 @@ from cartesia._types import (
     VoiceMetadata,
 )
 from cartesia.utils.retry import retry_on_connection_error, retry_on_connection_error_async
-from iterators import TimeoutIterator
-from websockets.sync.client import connect
 
 DEFAULT_MODEL_ID = "sonic-english"  # latest default model
 MULTILINGUAL_MODEL_ID = "sonic-multilingual"  # latest multilingual model
@@ -293,6 +294,7 @@ class _TTSContext:
         context_id: Optional[str] = None,
         duration: Optional[int] = None,
         language: Optional[str] = None,
+        add_timestamps: bool = False,
         _experimental_voice_controls: Optional[VoiceControls] = None,
     ) -> Generator[bytes, None, None]:
         """Send audio generation requests to the WebSocket and yield responses.
@@ -306,6 +308,7 @@ class _TTSContext:
             context_id: The context ID to use for the request. If not specified, a random context ID will be generated.
             duration: The duration of the audio in seconds.
             language: The language code for the audio request. This can only be used with `model_id = sonic-multilingual`
+            add_timestamps: Whether to return word-level timestamps.
             _experimental_voice_controls: Experimental voice controls for controlling speed and emotion.
                 Note: This is an experimental feature and may change rapidly in future releases.
 
@@ -340,6 +343,7 @@ class _TTSContext:
             },
             "context_id": self._context_id,
             "language": language,
+            "add_timestamps": add_timestamps,
         }
 
         if duration is not None:

cartesia-1.0.11/cartesia/version.py

@@ -0,0 +1 @@
+__version__ = "1.0.11"

{cartesia-1.0.9 → cartesia-1.0.11}/cartesia.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: cartesia
-Version: 1.0.9
+Version: 1.0.11
 Summary: The official Python library for the Cartesia API.
 Home-page: 
 Author: Cartesia, Inc.

{cartesia-1.0.9 → cartesia-1.0.11}/tests/test_deprecated.py

@@ -1,7 +1,7 @@
-import cartesia as Cartesia
+from packaging.version import Version
+
 import cartesia.version as version
 from cartesia.utils.deprecated import _DEPRECATED_FUNCTION_STATS
-from packaging.version import Version
 
 
 def test_deprecated_to_remove_by_version():

{cartesia-1.0.9 → cartesia-1.0.11}/tests/test_tts.py

@@ -14,6 +14,7 @@ from typing import AsyncGenerator, Generator, List
 
 import numpy as np
 import pytest
+
 from cartesia import AsyncCartesia, Cartesia
 from cartesia._types import VoiceControls, VoiceMetadata
 from cartesia.client import DEFAULT_MODEL_ID, MULTILINGUAL_MODEL_ID
@@ -25,6 +26,7 @@ RESOURCES_DIR = os.path.join(THISDIR, "resources")
 SAMPLE_VOICE = "Newsman"
 SAMPLE_VOICE_ID = "d46abd1d-2d02-43e8-819f-51fb652c1c61"
 EXPERIMENTAL_VOICE_CONTROLS = {"emotion": ["anger:high", "positivity:low"], "speed": "fastest"}
+EXPERIMENTAL_VOICE_CONTROLS_2 = {"speed": 0.4}
 
 logger = logging.getLogger(__name__)
 
@@ -102,7 +104,7 @@ def test_create_voice(client: Cartesia):
     assert voice in voices
 
 @pytest.mark.parametrize("stream", [True, False])
-@pytest.mark.parametrize("_experimental_voice_controls", [None, EXPERIMENTAL_VOICE_CONTROLS])
+@pytest.mark.parametrize("_experimental_voice_controls", [None, EXPERIMENTAL_VOICE_CONTROLS, EXPERIMENTAL_VOICE_CONTROLS_2])
 def test_sse_send(resources: _Resources, stream: bool, _experimental_voice_controls: VoiceControls):
     logger.info("Testing SSE send")
     client = resources.client
@@ -139,7 +141,7 @@ def test_sse_send_with_model_id(resources: _Resources, stream: bool):
         assert isinstance(out["audio"], bytes)
         
 @pytest.mark.parametrize("stream", [True, False])
-@pytest.mark.parametrize("_experimental_voice_controls", [None, EXPERIMENTAL_VOICE_CONTROLS])
+@pytest.mark.parametrize("_experimental_voice_controls", [None, EXPERIMENTAL_VOICE_CONTROLS, EXPERIMENTAL_VOICE_CONTROLS_2])
 def test_websocket_send(resources: _Resources, stream: bool, _experimental_voice_controls: VoiceControls):
     logger.info("Testing WebSocket send")
     client = resources.client
@@ -188,8 +190,7 @@ def test_websocket_send_timestamps(resources: _Resources, stream: bool):
     
     ws.close()
 
-        
-@pytest.mark.parametrize("_experimental_voice_controls", [None, EXPERIMENTAL_VOICE_CONTROLS])
+@pytest.mark.parametrize("_experimental_voice_controls", [None, EXPERIMENTAL_VOICE_CONTROLS, EXPERIMENTAL_VOICE_CONTROLS_2])
 def test_sse_send_context_manager(resources: _Resources, _experimental_voice_controls: VoiceControls):
     logger.info("Testing SSE send context manager")
     transcript = "Hello, world! I'\''m generating audio on Cartesia."
@@ -255,7 +256,7 @@ def test_websocket_send_context_manage_err(resources: _Resources):
         pass
 
 @pytest.mark.asyncio
-@pytest.mark.parametrize("_experimental_voice_controls", [None, EXPERIMENTAL_VOICE_CONTROLS])
+@pytest.mark.parametrize("_experimental_voice_controls", [None, EXPERIMENTAL_VOICE_CONTROLS, EXPERIMENTAL_VOICE_CONTROLS_2])
 async def test_async_sse_send( resources: _Resources, _experimental_voice_controls: VoiceControls):
     logger.info("Testing async SSE send")
     transcript = "Hello, world! I'\''m generating audio on Cartesia."
@@ -276,7 +277,7 @@ async def test_async_sse_send( resources: _Resources, _experimental_voice_contro
         await async_client.close()
         
 @pytest.mark.asyncio
-@pytest.mark.parametrize("_experimental_voice_controls", [None, EXPERIMENTAL_VOICE_CONTROLS])
+@pytest.mark.parametrize("_experimental_voice_controls", [None, EXPERIMENTAL_VOICE_CONTROLS, EXPERIMENTAL_VOICE_CONTROLS_2])
 async def test_async_websocket_send(resources: _Resources,  _experimental_voice_controls: VoiceControls):
     logger.info("Testing async WebSocket send")
     transcript = "Hello, world! I'\''m generating audio on Cartesia."
@@ -454,6 +455,28 @@ def test_sync_continuation_websocket_context_send():
             assert isinstance(out["audio"], bytes)
     finally:
         ws.close()
+        
+def test_sync_context_send_timestamps(resources: _Resources):
+    logger.info("Testing WebSocket send")
+    client = resources.client
+    transcripts = ["Hello, world!", "I'\''m generating audio on Cartesia."]
+
+    ws = client.tts.websocket()
+    ctx = ws.context()
+    output_generate = ctx.send(transcript=chunk_generator(transcripts), voice_id=SAMPLE_VOICE_ID, output_format={
+        "container": "raw",
+        "encoding": "pcm_f32le",
+        "sample_rate": 44100
+    }, model_id=DEFAULT_MODEL_ID, add_timestamps=True)
+
+    has_wordtimestamps = False
+    for out in output_generate:
+        has_wordtimestamps |= "word_timestamps" in out
+        _validate_schema(out)
+
+    assert has_wordtimestamps, "No word timestamps found"
+    
+    ws.close()
     
 @pytest.mark.asyncio
 async def test_continuation_websocket_context_send():

cartesia-1.0.9/cartesia/version.py

@@ -1 +0,0 @@
-__version__ = "1.0.9"