cartesia 2.0.2__tar.gz → 2.0.4__tar.gz

{cartesia-2.0.2 → cartesia-2.0.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: cartesia
-Version: 2.0.2
+Version: 2.0.4
 Summary: 
 Requires-Python: >=3.8,<4.0
 Classifier: Intended Audience :: Developers

{cartesia-2.0.2 → cartesia-2.0.4}/pyproject.toml

@@ -3,7 +3,7 @@ name = "cartesia"
 
 [tool.poetry]
 name = "cartesia"
-version = "2.0.2"
+version = "2.0.4"
 description = ""
 readme = "README.md"
 authors = []

{cartesia-2.0.2 → cartesia-2.0.4}/src/cartesia/__init__.py

@@ -29,6 +29,7 @@ from .tts import (
     FlushId,
     GenerationRequest,
     GenerationRequestParams,
+    ModelSpeed,
     Mp3OutputFormat,
     Mp3OutputFormatParams,
     NaturalSpecifier,
@@ -201,6 +202,7 @@ __all__ = [
     "MixVoiceSpecifierParams",
     "MixVoicesRequest",
     "MixVoicesRequestParams",
+    "ModelSpeed",
     "Mp3OutputFormat",
     "Mp3OutputFormatParams",
     "NaturalSpecifier",

{cartesia-2.0.2 → cartesia-2.0.4}/src/cartesia/core/client_wrapper.py

@@ -16,7 +16,7 @@ class BaseClientWrapper:
         headers: typing.Dict[str, str] = {
             "X-Fern-Language": "Python",
             "X-Fern-SDK-Name": "cartesia",
-            "X-Fern-SDK-Version": "2.0.2",
+            "X-Fern-SDK-Version": "2.0.4",
         }
         headers["X-API-Key"] = self.api_key
         headers["Cartesia-Version"] = "2024-11-13"

{cartesia-2.0.2 → cartesia-2.0.4}/src/cartesia/tts/__init__.py

@@ -7,6 +7,7 @@ from .types import (
     Emotion,
     FlushId,
     GenerationRequest,
+    ModelSpeed,
     Mp3OutputFormat,
     NaturalSpecifier,
     NumericalSpecifier,
@@ -94,6 +95,7 @@ __all__ = [
     "FlushId",
     "GenerationRequest",
     "GenerationRequestParams",
+    "ModelSpeed",
     "Mp3OutputFormat",
     "Mp3OutputFormatParams",
     "NaturalSpecifier",

{cartesia-2.0.2 → cartesia-2.0.4}/src/cartesia/tts/_websocket.py

@@ -154,7 +154,7 @@ class _TTSContext:
                             f"Error generating audio:\n{response_obj.error}"
                         )
                     if isinstance(response_obj, WebSocketResponse_Done):
-                        break
+                        return
                     if (
                         isinstance(response_obj, WebSocketResponse_Chunk)
                         or isinstance(response_obj, WebSocketResponse_Timestamps)
@@ -185,7 +185,7 @@ class _TTSContext:
                                 f"Error generating audio:\n{response_obj.error}"
                             )
                         if isinstance(response_obj, WebSocketResponse_Done):
-                            break
+                            return
                         if (
                             isinstance(response_obj, WebSocketResponse_Chunk)
                             or isinstance(response_obj, WebSocketResponse_Timestamps)
@@ -221,7 +221,7 @@ class _TTSContext:
                 if isinstance(response_obj, WebSocketResponse_Error):
                     raise RuntimeError(f"Error generating audio:\n{response_obj.error}")
                 if isinstance(response_obj, WebSocketResponse_Done):
-                    break
+                    return
                 yield self._websocket._convert_response(
                     response_obj, include_context_id=True
                 )

{cartesia-2.0.2 → cartesia-2.0.4}/src/cartesia/tts/client.py

@@ -5,6 +5,7 @@ from ..core.client_wrapper import SyncClientWrapper
 from .requests.tts_request_voice_specifier import TtsRequestVoiceSpecifierParams
 from .requests.output_format import OutputFormatParams
 from .types.supported_language import SupportedLanguage
+from .types.model_speed import ModelSpeed
 from ..core.request_options import RequestOptions
 from ..core.serialization import convert_and_respect_annotation_metadata
 from json.decoder import JSONDecodeError
@@ -32,7 +33,7 @@ class TtsClient:
         output_format: OutputFormatParams,
         language: typing.Optional[SupportedLanguage] = OMIT,
         duration: typing.Optional[float] = OMIT,
-        text_cfg: typing.Optional[float] = OMIT,
+        speed: typing.Optional[ModelSpeed] = OMIT,
         request_options: typing.Optional[RequestOptions] = None,
     ) -> typing.Iterator[bytes]:
         """
@@ -53,12 +54,7 @@ class TtsClient:
             The maximum duration of the audio in seconds. You do not usually need to specify this.
             If the duration is not appropriate for the length of the transcript, the output audio may be truncated.
 
-        text_cfg : typing.Optional[float]
-            The text [classifier-free guidance](https://arxiv.org/abs/2207.12598) value for the request.
-
-            Higher values causes the model to attend more to the text but speed up the generation. Lower values reduce the speaking rate but can increase the risk of hallucinations. The default value is `3.0`. For a slower speaking rate, we recommend values between `2.0` and `3.0`. Values are supported between `1.5` and `3.0`.
-
-            This parameter is only supported for `sonic-2` models.
+        speed : typing.Optional[ModelSpeed]
 
         request_options : typing.Optional[RequestOptions]
             Request-specific configuration. You can pass in configuration such as `chunk_size`, and more to customize the request and response.
@@ -100,7 +96,7 @@ class TtsClient:
                     object_=output_format, annotation=OutputFormatParams, direction="write"
                 ),
                 "duration": duration,
-                "text_cfg": text_cfg,
+                "speed": speed,
             },
             request_options=request_options,
             omit=OMIT,
@@ -126,7 +122,7 @@ class TtsClient:
         output_format: OutputFormatParams,
         language: typing.Optional[SupportedLanguage] = OMIT,
         duration: typing.Optional[float] = OMIT,
-        text_cfg: typing.Optional[float] = OMIT,
+        speed: typing.Optional[ModelSpeed] = OMIT,
         request_options: typing.Optional[RequestOptions] = None,
     ) -> typing.Iterator[WebSocketResponse]:
         """
@@ -147,12 +143,7 @@ class TtsClient:
             The maximum duration of the audio in seconds. You do not usually need to specify this.
             If the duration is not appropriate for the length of the transcript, the output audio may be truncated.
 
-        text_cfg : typing.Optional[float]
-            The text [classifier-free guidance](https://arxiv.org/abs/2207.12598) value for the request.
-
-            Higher values causes the model to attend more to the text but speed up the generation. Lower values reduce the speaking rate but can increase the risk of hallucinations. The default value is `3.0`. For a slower speaking rate, we recommend values between `2.0` and `3.0`. Values are supported between `1.5` and `3.0`.
-
-            This parameter is only supported for `sonic-2` models.
+        speed : typing.Optional[ModelSpeed]
 
         request_options : typing.Optional[RequestOptions]
             Request-specific configuration.
@@ -196,7 +187,7 @@ class TtsClient:
                     object_=output_format, annotation=OutputFormatParams, direction="write"
                 ),
                 "duration": duration,
-                "text_cfg": text_cfg,
+                "speed": speed,
             },
             request_options=request_options,
             omit=OMIT,
@@ -236,7 +227,7 @@ class AsyncTtsClient:
         output_format: OutputFormatParams,
         language: typing.Optional[SupportedLanguage] = OMIT,
         duration: typing.Optional[float] = OMIT,
-        text_cfg: typing.Optional[float] = OMIT,
+        speed: typing.Optional[ModelSpeed] = OMIT,
         request_options: typing.Optional[RequestOptions] = None,
     ) -> typing.AsyncIterator[bytes]:
         """
@@ -257,12 +248,7 @@ class AsyncTtsClient:
             The maximum duration of the audio in seconds. You do not usually need to specify this.
             If the duration is not appropriate for the length of the transcript, the output audio may be truncated.
 
-        text_cfg : typing.Optional[float]
-            The text [classifier-free guidance](https://arxiv.org/abs/2207.12598) value for the request.
-
-            Higher values causes the model to attend more to the text but speed up the generation. Lower values reduce the speaking rate but can increase the risk of hallucinations. The default value is `3.0`. For a slower speaking rate, we recommend values between `2.0` and `3.0`. Values are supported between `1.5` and `3.0`.
-
-            This parameter is only supported for `sonic-2` models.
+        speed : typing.Optional[ModelSpeed]
 
         request_options : typing.Optional[RequestOptions]
             Request-specific configuration. You can pass in configuration such as `chunk_size`, and more to customize the request and response.
@@ -312,7 +298,7 @@ class AsyncTtsClient:
                     object_=output_format, annotation=OutputFormatParams, direction="write"
                 ),
                 "duration": duration,
-                "text_cfg": text_cfg,
+                "speed": speed,
             },
             request_options=request_options,
             omit=OMIT,
@@ -338,7 +324,7 @@ class AsyncTtsClient:
         output_format: OutputFormatParams,
         language: typing.Optional[SupportedLanguage] = OMIT,
         duration: typing.Optional[float] = OMIT,
-        text_cfg: typing.Optional[float] = OMIT,
+        speed: typing.Optional[ModelSpeed] = OMIT,
         request_options: typing.Optional[RequestOptions] = None,
     ) -> typing.AsyncIterator[WebSocketResponse]:
         """
@@ -359,12 +345,7 @@ class AsyncTtsClient:
             The maximum duration of the audio in seconds. You do not usually need to specify this.
             If the duration is not appropriate for the length of the transcript, the output audio may be truncated.
 
-        text_cfg : typing.Optional[float]
-            The text [classifier-free guidance](https://arxiv.org/abs/2207.12598) value for the request.
-
-            Higher values causes the model to attend more to the text but speed up the generation. Lower values reduce the speaking rate but can increase the risk of hallucinations. The default value is `3.0`. For a slower speaking rate, we recommend values between `2.0` and `3.0`. Values are supported between `1.5` and `3.0`.
-
-            This parameter is only supported for `sonic-2` models.
+        speed : typing.Optional[ModelSpeed]
 
         request_options : typing.Optional[RequestOptions]
             Request-specific configuration.
@@ -416,7 +397,7 @@ class AsyncTtsClient:
                     object_=output_format, annotation=OutputFormatParams, direction="write"
                 ),
                 "duration": duration,
-                "text_cfg": text_cfg,
+                "speed": speed,
             },
             request_options=request_options,
             omit=OMIT,

{cartesia-2.0.2 → cartesia-2.0.4}/src/cartesia/tts/requests/generation_request.py

@@ -6,6 +6,7 @@ from .tts_request_voice_specifier import TtsRequestVoiceSpecifierParams
 import typing_extensions
 from ..types.supported_language import SupportedLanguage
 from .web_socket_raw_output_format import WebSocketRawOutputFormatParams
+from ..types.model_speed import ModelSpeed
 from ..types.context_id import ContextId
 from ...core.serialization import FieldMetadata
 
@@ -30,15 +31,7 @@ class GenerationRequestParams(typing_extensions.TypedDict):
     If the duration is not appropriate for the length of the transcript, the output audio may be truncated.
     """
 
-    text_cfg: typing_extensions.NotRequired[float]
-    """
-    The text [classifier-free guidance](https://arxiv.org/abs/2207.12598) value for the request.
-    
-    Higher values causes the model to attend more to the text but speed up the generation. Lower values reduce the speaking rate but can increase the risk of hallucinations. The default value is `3.0`. For a slower speaking rate, we recommend values between `2.0` and `3.0`. Values are supported between `1.5` and `3.0`.
-    
-    This parameter is only supported for `sonic-2` models.
-    """
-
+    speed: typing_extensions.NotRequired[ModelSpeed]
     context_id: typing_extensions.NotRequired[ContextId]
     continue_: typing_extensions.NotRequired[typing_extensions.Annotated[bool, FieldMetadata(alias="continue")]]
     """

{cartesia-2.0.2 → cartesia-2.0.4}/src/cartesia/tts/requests/tts_request.py

@@ -5,6 +5,7 @@ from .tts_request_voice_specifier import TtsRequestVoiceSpecifierParams
 import typing_extensions
 from ..types.supported_language import SupportedLanguage
 from .output_format import OutputFormatParams
+from ..types.model_speed import ModelSpeed
 
 
 class TtsRequestParams(typing_extensions.TypedDict):
@@ -23,11 +24,4 @@ class TtsRequestParams(typing_extensions.TypedDict):
     If the duration is not appropriate for the length of the transcript, the output audio may be truncated.
     """
 
-    text_cfg: typing_extensions.NotRequired[float]
-    """
-    The text [classifier-free guidance](https://arxiv.org/abs/2207.12598) value for the request.
-    
-    Higher values causes the model to attend more to the text but speed up the generation. Lower values reduce the speaking rate but can increase the risk of hallucinations. The default value is `3.0`. For a slower speaking rate, we recommend values between `2.0` and `3.0`. Values are supported between `1.5` and `3.0`.
-    
-    This parameter is only supported for `sonic-2` models.
-    """
+    speed: typing_extensions.NotRequired[ModelSpeed]

{cartesia-2.0.2 → cartesia-2.0.4}/src/cartesia/tts/requests/web_socket_tts_request.py

@@ -5,6 +5,7 @@ import typing_extensions
 from .output_format import OutputFormatParams
 from .tts_request_voice_specifier import TtsRequestVoiceSpecifierParams
 from ...core.serialization import FieldMetadata
+from ..types.model_speed import ModelSpeed
 
 
 class WebSocketTtsRequestParams(typing_extensions.TypedDict):
@@ -24,4 +25,4 @@ class WebSocketTtsRequestParams(typing_extensions.TypedDict):
     continue_: typing_extensions.NotRequired[typing_extensions.Annotated[bool, FieldMetadata(alias="continue")]]
     context_id: typing_extensions.NotRequired[str]
     max_buffer_delay_ms: typing_extensions.NotRequired[int]
-    text_cfg: typing_extensions.NotRequired[float]
+    speed: typing_extensions.NotRequired[ModelSpeed]

{cartesia-2.0.2 → cartesia-2.0.4}/src/cartesia/tts/types/__init__.py

@@ -6,6 +6,7 @@ from .controls import Controls
 from .emotion import Emotion
 from .flush_id import FlushId
 from .generation_request import GenerationRequest
+from .model_speed import ModelSpeed
 from .mp_3_output_format import Mp3OutputFormat
 from .natural_specifier import NaturalSpecifier
 from .numerical_specifier import NumericalSpecifier
@@ -50,6 +51,7 @@ __all__ = [
     "Emotion",
     "FlushId",
     "GenerationRequest",
+    "ModelSpeed",
     "Mp3OutputFormat",
     "NaturalSpecifier",
     "NumericalSpecifier",

{cartesia-2.0.2 → cartesia-2.0.4}/src/cartesia/tts/types/generation_request.py

@@ -6,6 +6,7 @@ import typing
 from .tts_request_voice_specifier import TtsRequestVoiceSpecifier
 from .supported_language import SupportedLanguage
 from .web_socket_raw_output_format import WebSocketRawOutputFormat
+from .model_speed import ModelSpeed
 from .context_id import ContextId
 import typing_extensions
 from ...core.serialization import FieldMetadata
@@ -32,15 +33,7 @@ class GenerationRequest(UniversalBaseModel):
     If the duration is not appropriate for the length of the transcript, the output audio may be truncated.
     """
 
-    text_cfg: typing.Optional[float] = pydantic.Field(default=None)
-    """
-    The text [classifier-free guidance](https://arxiv.org/abs/2207.12598) value for the request.
-    
-    Higher values causes the model to attend more to the text but speed up the generation. Lower values reduce the speaking rate but can increase the risk of hallucinations. The default value is `3.0`. For a slower speaking rate, we recommend values between `2.0` and `3.0`. Values are supported between `1.5` and `3.0`.
-    
-    This parameter is only supported for `sonic-2` models.
-    """
-
+    speed: typing.Optional[ModelSpeed] = None
     context_id: typing.Optional[ContextId] = None
     continue_: typing_extensions.Annotated[typing.Optional[bool], FieldMetadata(alias="continue")] = pydantic.Field(
         default=None

cartesia-2.0.4/src/cartesia/tts/types/model_speed.py

@@ -0,0 +1,5 @@
+# This file was auto-generated by Fern from our API Definition.
+
+import typing
+
+ModelSpeed = typing.Union[typing.Literal["slow", "normal", "fast"], typing.Any]

{cartesia-2.0.2 → cartesia-2.0.4}/src/cartesia/tts/types/tts_request.py

@@ -6,6 +6,7 @@ from .tts_request_voice_specifier import TtsRequestVoiceSpecifier
 import typing
 from .supported_language import SupportedLanguage
 from .output_format import OutputFormat
+from .model_speed import ModelSpeed
 from ...core.pydantic_utilities import IS_PYDANTIC_V2
 
 
@@ -25,14 +26,7 @@ class TtsRequest(UniversalBaseModel):
     If the duration is not appropriate for the length of the transcript, the output audio may be truncated.
     """
 
-    text_cfg: typing.Optional[float] = pydantic.Field(default=None)
-    """
-    The text [classifier-free guidance](https://arxiv.org/abs/2207.12598) value for the request.
-    
-    Higher values causes the model to attend more to the text but speed up the generation. Lower values reduce the speaking rate but can increase the risk of hallucinations. The default value is `3.0`. For a slower speaking rate, we recommend values between `2.0` and `3.0`. Values are supported between `1.5` and `3.0`.
-    
-    This parameter is only supported for `sonic-2` models.
-    """
+    speed: typing.Optional[ModelSpeed] = None
 
     if IS_PYDANTIC_V2:
         model_config: typing.ClassVar[pydantic.ConfigDict] = pydantic.ConfigDict(extra="allow", frozen=True)  # type: ignore # Pydantic v2

{cartesia-2.0.2 → cartesia-2.0.4}/src/cartesia/tts/types/web_socket_tts_request.py

@@ -7,6 +7,7 @@ from .output_format import OutputFormat
 from .tts_request_voice_specifier import TtsRequestVoiceSpecifier
 import typing_extensions
 from ...core.serialization import FieldMetadata
+from .model_speed import ModelSpeed
 from ...core.pydantic_utilities import IS_PYDANTIC_V2
 
 
@@ -27,7 +28,7 @@ class WebSocketTtsRequest(UniversalBaseModel):
     continue_: typing_extensions.Annotated[typing.Optional[bool], FieldMetadata(alias="continue")] = None
     context_id: typing.Optional[str] = None
     max_buffer_delay_ms: typing.Optional[int] = None
-    text_cfg: typing.Optional[float] = None
+    speed: typing.Optional[ModelSpeed] = None
 
     if IS_PYDANTIC_V2:
         model_config: typing.ClassVar[pydantic.ConfigDict] = pydantic.ConfigDict(extra="allow", frozen=True)  # type: ignore # Pydantic v2

{cartesia-2.0.2 → cartesia-2.0.4}/src/cartesia/voices/client.py

@@ -14,8 +14,8 @@ from ..core.api_error import ApiError
 from .. import core
 from ..tts.types.supported_language import SupportedLanguage
 from .types.clone_mode import CloneMode
-from .types.voice_metadata import VoiceMetadata
 from .types.voice_id import VoiceId
+from .types.voice_metadata import VoiceMetadata
 from ..core.jsonable_encoder import jsonable_encoder
 from .types.localize_target_language import LocalizeTargetLanguage
 from .types.gender import Gender
@@ -150,7 +150,7 @@ class VoicesClient:
         mode: CloneMode,
         description: typing.Optional[str] = OMIT,
         enhance: typing.Optional[bool] = OMIT,
-        transcript: typing.Optional[str] = OMIT,
+        base_voice_id: typing.Optional[VoiceId] = OMIT,
         request_options: typing.Optional[RequestOptions] = None,
     ) -> VoiceMetadata:
         """
@@ -182,11 +182,11 @@ class VoicesClient:
 
 
         enhance : typing.Optional[bool]
-            Whether to enhance the clip to improve its quality before cloning. Useful if the clip has background noise.
+            Whether to apply AI enhancements to the clip to reduce background noise. This leads to cleaner generated speech at the cost of reduced similarity to the source clip.
 
 
-        transcript : typing.Optional[str]
-            Optional transcript of the words spoken in the audio clip. Only used for similarity mode.
+        base_voice_id : typing.Optional[VoiceId]
+            Optional base voice ID that the cloned voice is derived from.
 
 
         request_options : typing.Optional[RequestOptions]
@@ -208,7 +208,6 @@ class VoicesClient:
             description="Copied from Cartesia docs",
             mode="stability",
             language="en",
-            enhance=True,
         )
         """
         _response = self._client_wrapper.httpx_client.request(
@@ -220,7 +219,7 @@ class VoicesClient:
                 "language": language,
                 "mode": mode,
                 "enhance": enhance,
-                "transcript": transcript,
+                "base_voice_id": base_voice_id,
             },
             files={
                 "clip": clip,
@@ -719,7 +718,7 @@ class AsyncVoicesClient:
         mode: CloneMode,
         description: typing.Optional[str] = OMIT,
         enhance: typing.Optional[bool] = OMIT,
-        transcript: typing.Optional[str] = OMIT,
+        base_voice_id: typing.Optional[VoiceId] = OMIT,
         request_options: typing.Optional[RequestOptions] = None,
     ) -> VoiceMetadata:
         """
@@ -751,11 +750,11 @@ class AsyncVoicesClient:
 
 
         enhance : typing.Optional[bool]
-            Whether to enhance the clip to improve its quality before cloning. Useful if the clip has background noise.
+            Whether to apply AI enhancements to the clip to reduce background noise. This leads to cleaner generated speech at the cost of reduced similarity to the source clip.
 
 
-        transcript : typing.Optional[str]
-            Optional transcript of the words spoken in the audio clip. Only used for similarity mode.
+        base_voice_id : typing.Optional[VoiceId]
+            Optional base voice ID that the cloned voice is derived from.
 
 
         request_options : typing.Optional[RequestOptions]
@@ -782,7 +781,6 @@ class AsyncVoicesClient:
                 description="Copied from Cartesia docs",
                 mode="stability",
                 language="en",
-                enhance=True,
             )
 
 
@@ -797,7 +795,7 @@ class AsyncVoicesClient:
                 "language": language,
                 "mode": mode,
                 "enhance": enhance,
-                "transcript": transcript,
+                "base_voice_id": base_voice_id,
             },
             files={
                 "clip": clip,