cartesia 2.0.0b1__py3-none-any.whl → 2.0.0b7__py3-none-any.whl

cartesia/__init__.py

@@ -121,13 +121,14 @@ from .voices import (
     EmbeddingSpecifier,
     EmbeddingSpecifierParams,
     Gender,
+    GenderPresentation,
+    GetVoicesResponse,
+    GetVoicesResponseParams,
     IdSpecifier,
     IdSpecifierParams,
     LocalizeDialect,
     LocalizeDialectParams,
     LocalizeEnglishDialect,
-    LocalizePortugueseDialect,
-    LocalizeSpanishDialect,
     LocalizeTargetLanguage,
     LocalizeVoiceRequest,
     LocalizeVoiceRequestParams,
@@ -138,6 +139,7 @@ from .voices import (
     UpdateVoiceRequest,
     UpdateVoiceRequestParams,
     Voice,
+    VoiceExpandOptions,
     VoiceId,
     VoiceMetadata,
     VoiceMetadataParams,
@@ -175,15 +177,16 @@ __all__ = [
     "FilePurpose",
     "FlushId",
     "Gender",
+    "GenderPresentation",
     "GenerationRequest",
     "GenerationRequestParams",
+    "GetVoicesResponse",
+    "GetVoicesResponseParams",
     "IdSpecifier",
     "IdSpecifierParams",
     "LocalizeDialect",
     "LocalizeDialectParams",
     "LocalizeEnglishDialect",
-    "LocalizePortugueseDialect",
-    "LocalizeSpanishDialect",
     "LocalizeTargetLanguage",
     "LocalizeVoiceRequest",
     "LocalizeVoiceRequestParams",
@@ -235,6 +238,7 @@ __all__ = [
     "UpdateVoiceRequest",
     "UpdateVoiceRequestParams",
     "Voice",
+    "VoiceExpandOptions",
     "VoiceId",
     "VoiceMetadata",
     "VoiceMetadataParams",

cartesia/base_client.py

@@ -5,14 +5,12 @@ from .environment import CartesiaEnvironment
 import httpx
 from .core.client_wrapper import SyncClientWrapper
 from .api_status.client import ApiStatusClient
-from .datasets.client import DatasetsClient
 from .infill.client import InfillClient
 from .tts.client import TtsClient
 from .voice_changer.client import VoiceChangerClient
 from .voices.client import VoicesClient
 from .core.client_wrapper import AsyncClientWrapper
 from .api_status.client import AsyncApiStatusClient
-from .datasets.client import AsyncDatasetsClient
 from .infill.client import AsyncInfillClient
 from .tts.client import AsyncTtsClient
 from .voice_changer.client import AsyncVoiceChangerClient
@@ -78,7 +76,6 @@ class BaseCartesia:
             timeout=_defaulted_timeout,
         )
         self.api_status = ApiStatusClient(client_wrapper=self._client_wrapper)
-        self.datasets = DatasetsClient(client_wrapper=self._client_wrapper)
         self.infill = InfillClient(client_wrapper=self._client_wrapper)
         self.tts = TtsClient(client_wrapper=self._client_wrapper)
         self.voice_changer = VoiceChangerClient(client_wrapper=self._client_wrapper)
@@ -144,7 +141,6 @@ class AsyncBaseCartesia:
             timeout=_defaulted_timeout,
         )
         self.api_status = AsyncApiStatusClient(client_wrapper=self._client_wrapper)
-        self.datasets = AsyncDatasetsClient(client_wrapper=self._client_wrapper)
         self.infill = AsyncInfillClient(client_wrapper=self._client_wrapper)
         self.tts = AsyncTtsClient(client_wrapper=self._client_wrapper)
         self.voice_changer = AsyncVoiceChangerClient(client_wrapper=self._client_wrapper)

cartesia/core/__init__.py

@@ -6,6 +6,7 @@ from .datetime_utils import serialize_datetime
 from .file import File, convert_file_dict_to_httpx_tuples, with_content_type
 from .http_client import AsyncHttpClient, HttpClient
 from .jsonable_encoder import jsonable_encoder
+from .pagination import AsyncPager, SyncPager
 from .pydantic_utilities import (
     IS_PYDANTIC_V2,
     UniversalBaseModel,
@@ -24,6 +25,7 @@ __all__ = [
     "ApiError",
     "AsyncClientWrapper",
     "AsyncHttpClient",
+    "AsyncPager",
     "BaseClientWrapper",
     "FieldMetadata",
     "File",
@@ -31,6 +33,7 @@ __all__ = [
     "IS_PYDANTIC_V2",
     "RequestOptions",
     "SyncClientWrapper",
+    "SyncPager",
     "UniversalBaseModel",
     "UniversalRootModel",
     "convert_and_respect_annotation_metadata",

cartesia/core/client_wrapper.py

@@ -16,10 +16,10 @@ class BaseClientWrapper:
         headers: typing.Dict[str, str] = {
             "X-Fern-Language": "Python",
             "X-Fern-SDK-Name": "cartesia",
-            "X-Fern-SDK-Version": "2.0.0b1",
+            "X-Fern-SDK-Version": "2.0.0b7",
         }
         headers["X-API-Key"] = self.api_key
-        headers["Cartesia-Version"] = "2024-06-10"
+        headers["Cartesia-Version"] = "2024-11-13"
         return headers
 
     def get_base_url(self) -> str:

cartesia/core/pagination.py

@@ -0,0 +1,88 @@
+# This file was auto-generated by Fern from our API Definition.
+
+import typing
+
+from typing_extensions import Self
+
+import pydantic
+
+# Generic to represent the underlying type of the results within a page
+T = typing.TypeVar("T")
+
+
+# SDKs implement a Page ABC per-pagination request, the endpoint then returns a pager that wraps this type
+# for example, an endpoint will return SyncPager[UserPage] where UserPage implements the Page ABC. ex:
+#
+# SyncPager<InnerListType>(
+#     has_next=response.list_metadata.after is not None,
+#     items=response.data,
+#     # This should be the outer function that returns the SyncPager again
+#     get_next=lambda: list(..., cursor: response.cursor) (or list(..., offset: offset + 1))
+# )
+class BasePage(pydantic.BaseModel, typing.Generic[T]):
+    has_next: bool
+    items: typing.Optional[typing.List[T]]
+
+
+class SyncPage(BasePage[T], typing.Generic[T]):
+    get_next: typing.Optional[typing.Callable[[], typing.Optional[Self]]]
+
+
+class AsyncPage(BasePage[T], typing.Generic[T]):
+    get_next: typing.Optional[typing.Callable[[], typing.Awaitable[typing.Optional[Self]]]]
+
+
+# ----------------------------
+
+
+class SyncPager(SyncPage[T], typing.Generic[T]):
+    # Here we type ignore the iterator to avoid a mypy error
+    # caused by the type conflict with Pydanitc's __iter__ method
+    # brought in by extending the base model
+    def __iter__(self) -> typing.Iterator[T]:  # type: ignore
+        for page in self.iter_pages():
+            if page.items is not None:
+                for item in page.items:
+                    yield item
+
+    def iter_pages(self) -> typing.Iterator[SyncPage[T]]:
+        page: typing.Union[SyncPager[T], None] = self
+        while True:
+            if page is not None:
+                yield page
+                if page.has_next and page.get_next is not None:
+                    page = page.get_next()
+                    if page is None or page.items is None or len(page.items) == 0:
+                        return
+                else:
+                    return
+            else:
+                return
+
+    def next_page(self) -> typing.Optional[SyncPage[T]]:
+        return self.get_next() if self.get_next is not None else None
+
+
+class AsyncPager(AsyncPage[T], typing.Generic[T]):
+    async def __aiter__(self) -> typing.AsyncIterator[T]:  # type: ignore
+        async for page in self.iter_pages():
+            if page.items is not None:
+                for item in page.items:
+                    yield item
+
+    async def iter_pages(self) -> typing.AsyncIterator[AsyncPage[T]]:
+        page: typing.Union[AsyncPager[T], None] = self
+        while True:
+            if page is not None:
+                yield page
+                if page is not None and page.has_next and page.get_next is not None:
+                    page = await page.get_next()
+                    if page is None or page.items is None or len(page.items) == 0:
+                        return
+                else:
+                    return
+            else:
+                return
+
+    async def next_page(self) -> typing.Optional[AsyncPage[T]]:
+        return await self.get_next() if self.get_next is not None else None

cartesia/infill/client.py

@@ -42,7 +42,7 @@ class InfillClient:
 
         **The cost is 1 credit per character of the infill text plus a fixed cost of 300 credits.**
 
-        Only the `sonic-preview` model is supported for infill at this time.
+        Infilling is only available on `sonic-2` at this time.
 
         At least one of `left_audio` or `right_audio` must be provided.
 
@@ -117,7 +117,7 @@ class InfillClient:
             api_key="YOUR_API_KEY",
         )
         client.infill.bytes(
-            model_id="sonic-preview",
+            model_id="sonic-2",
             language="en",
             transcript="middle segment",
             voice_id="694f9389-aac1-45b6-b726-9d9369183238",
@@ -189,7 +189,7 @@ class AsyncInfillClient:
 
         **The cost is 1 credit per character of the infill text plus a fixed cost of 300 credits.**
 
-        Only the `sonic-preview` model is supported for infill at this time.
+        Infilling is only available on `sonic-2` at this time.
 
         At least one of `left_audio` or `right_audio` must be provided.
 
@@ -269,7 +269,7 @@ class AsyncInfillClient:
 
         async def main() -> None:
             await client.infill.bytes(
-                model_id="sonic-preview",
+                model_id="sonic-2",
                 language="en",
                 transcript="middle segment",
                 voice_id="694f9389-aac1-45b6-b726-9d9369183238",

cartesia/tts/_async_websocket.py

@@ -17,6 +17,7 @@ from cartesia.tts.types import (
     WebSocketResponse_FlushDone,
     WebSocketTtsOutput,
     WordTimestamps,
+    PhonemeTimestamps,
 )
 
 from ..core.pydantic_utilities import parse_obj_as
@@ -67,6 +68,7 @@ class _AsyncTTSContext:
         language: Optional[str] = None,
         stream: bool = True,
         add_timestamps: bool = False,
+        add_phoneme_timestamps: bool = False,
         continue_: bool = False,
         flush: bool = False,
     ) -> None:
@@ -102,6 +104,8 @@ class _AsyncTTSContext:
             request_body["stream"] = stream
         if add_timestamps:
             request_body["add_timestamps"] = add_timestamps
+        if add_phoneme_timestamps:
+            request_body["add_phoneme_timestamps"] = add_phoneme_timestamps
         if continue_:
             request_body["continue"] = continue_
         if flush:
@@ -229,6 +233,11 @@ class _AsyncTTSContext:
         finally:
             self._close()
 
+    async def cancel(self):
+        """Cancel the context. This will stop the generation of audio for this context."""
+        await self._websocket.websocket.send_json({"context_id": self._context_id, "cancel": True})
+        self._close()
+
     def _close(self) -> None:
         """Closes the context. Automatically called when a done message is received for this context."""
         self._websocket._remove_context(self._context_id)
@@ -297,7 +306,26 @@ class AsyncTtsWebsocket(TtsWebsocket):
             try:
                 self.websocket = await session.ws_connect(url)
             except Exception as e:
-                raise RuntimeError(f"Failed to connect to WebSocket at {url}. {e}")
+                # Extract status code if available
+                status_code = None
+                error_message = str(e)
+                
+                if hasattr(e, 'status') and e.status is not None:
+                    status_code = e.status
+                
+                    # Create a meaningful error message based on status code
+                    if status_code == 402:
+                        error_message = "Payment required. Your API key may have insufficient credits or permissions."
+                    elif status_code == 401:
+                        error_message = "Unauthorized. Please check your API key."
+                    elif status_code == 403:
+                        error_message = "Forbidden. You don't have permission to access this resource."
+                    elif status_code == 404:
+                        error_message = "Not found. The requested resource doesn't exist."
+                    
+                    raise RuntimeError(f"Failed to connect to WebSocket.\nStatus: {status_code}. Error message: {error_message}")
+                else:
+                    raise RuntimeError(f"Failed to connect to WebSocket at {url}. {e}")
 
     def _is_websocket_closed(self):
         return self.websocket.closed
@@ -338,6 +366,7 @@ class AsyncTtsWebsocket(TtsWebsocket):
         language: Optional[str] = None,
         stream: bool = True,
         add_timestamps: bool = False,
+        add_phoneme_timestamps: bool = False,
     ):
         """See :meth:`_WebSocket.send` for details."""
         if context_id is None:
@@ -355,6 +384,7 @@ class AsyncTtsWebsocket(TtsWebsocket):
             language=language,
             continue_=False,
             add_timestamps=add_timestamps,
+            add_phoneme_timestamps=add_phoneme_timestamps,
         )
 
         generator = ctx.receive()
@@ -366,6 +396,9 @@ class AsyncTtsWebsocket(TtsWebsocket):
         words: typing.List[str] = []
         start: typing.List[float] = []
         end: typing.List[float] = []
+        phonemes: typing.List[str] = []
+        phoneme_start: typing.List[float] = []
+        phoneme_end: typing.List[float] = []
         async for chunk in generator:
             if chunk.audio is not None:
                 chunks.append(chunk.audio)
@@ -374,6 +407,11 @@ class AsyncTtsWebsocket(TtsWebsocket):
                     words.extend(chunk.word_timestamps.words)
                     start.extend(chunk.word_timestamps.start)
                     end.extend(chunk.word_timestamps.end)
+            if add_phoneme_timestamps and chunk.phoneme_timestamps is not None:
+                if chunk.phoneme_timestamps is not None:
+                    phonemes.extend(chunk.phoneme_timestamps.phonemes)
+                    phoneme_start.extend(chunk.phoneme_timestamps.start)
+                    phoneme_end.extend(chunk.phoneme_timestamps.end)
 
         return WebSocketTtsOutput(
             audio=b"".join(chunks),  # type: ignore
@@ -387,6 +425,15 @@ class AsyncTtsWebsocket(TtsWebsocket):
                 if add_timestamps
                 else None
             ),
+            phoneme_timestamps=(
+                PhonemeTimestamps(
+                    phonemes=phonemes,
+                    start=phoneme_start,
+                    end=phoneme_end,
+                )
+                if add_phoneme_timestamps
+                else None
+            ),
         )
 
     async def _process_responses(self):

cartesia/tts/_websocket.py

@@ -26,6 +26,7 @@ from cartesia.tts.types import (
     WebSocketResponse_Timestamps,
     WebSocketTtsOutput,
     WordTimestamps,
+    PhonemeTimestamps,
 )
 
 from ..core.pydantic_utilities import parse_obj_as
@@ -58,7 +59,7 @@ class _TTSContext:
         self,
         *,
         model_id: str,
-        transcript: str,
+        transcript: typing.Generator[str, None, None],
         output_format: OutputFormatParams,
         voice: TtsRequestVoiceSpecifierParams,
         context_id: Optional[str] = None,
@@ -235,7 +236,7 @@ class TtsWebsocket:
     Usage:
         >>> ws = client.tts.websocket()
         >>> generation_request = GenerationRequest(
-        ...     model_id="sonic-english",
+        ...     model_id="sonic-2",
         ...     transcript="Hello world!",
         ...     voice_embedding=embedding
         ...     output_format={"container": "raw", "encoding": "pcm_f32le", "sample_rate": 44100}
@@ -281,7 +282,26 @@ class TtsWebsocket:
                     f"{self.ws_url}/{route}?api_key={self.api_key}&cartesia_version={self.cartesia_version}"
                 )
             except Exception as e:
-                raise RuntimeError(f"Failed to connect to WebSocket. {e}")
+                # Extract status code if available
+                status_code = None
+                error_message = str(e)
+                
+                if hasattr(e, 'status') and e.status is not None:
+                    status_code = e.status
+                
+                    # Create a meaningful error message based on status code
+                    if status_code == 402:
+                        error_message = "Payment required. Your API key may have insufficient credits or permissions."
+                    elif status_code == 401:
+                        error_message = "Unauthorized. Please check your API key."
+                    elif status_code == 403:
+                        error_message = "Forbidden. You don't have permission to access this resource."
+                    elif status_code == 404:
+                        error_message = "Not found. The requested resource doesn't exist."
+                    
+                    raise RuntimeError(f"Failed to connect to WebSocket.\nStatus: {status_code}. Error message: {error_message}")
+                else:
+                    raise RuntimeError(f"Failed to connect to WebSocket. {e}")
 
     def _is_websocket_closed(self):
         return self.websocket.socket.fileno() == -1
@@ -310,6 +330,8 @@ class TtsWebsocket:
             out["audio"] = base64.b64decode(response.data)
         elif isinstance(response, WebSocketResponse_Timestamps):
             out["word_timestamps"] = response.word_timestamps  # type: ignore
+        elif isinstance(response, WebSocketResponse_PhonemeTimestamps):
+            out["phoneme_timestamps"] = response.phoneme_timestamps  # type: ignore
         elif include_flush_id and isinstance(response, WebSocketResponse_FlushDone):
             out["flush_done"] = response.flush_done  # type: ignore
             out["flush_id"] = response.flush_id  # type: ignore
@@ -331,6 +353,7 @@ class TtsWebsocket:
         language: Optional[str] = None,
         stream: bool = True,
         add_timestamps: bool = False,
+        add_phoneme_timestamps: bool = False,
     ):
         """Send a request to the WebSocket to generate audio.
 
@@ -360,6 +383,7 @@ class TtsWebsocket:
             "language": language,
             "stream": stream,
             "add_timestamps": add_timestamps,
+            "add_phoneme_timestamps": add_phoneme_timestamps,
         }
         generator = self._websocket_generator(request_body)
 
@@ -370,6 +394,9 @@ class TtsWebsocket:
         words: typing.List[str] = []
         start: typing.List[float] = []
         end: typing.List[float] = []
+        phonemes: typing.List[str] = []
+        phoneme_start: typing.List[float] = []
+        phoneme_end: typing.List[float] = []
         for chunk in generator:
             if chunk.audio is not None:
                 chunks.append(chunk.audio)
@@ -378,6 +405,11 @@ class TtsWebsocket:
                     words.extend(chunk.word_timestamps.words)
                     start.extend(chunk.word_timestamps.start)
                     end.extend(chunk.word_timestamps.end)
+            if add_phoneme_timestamps and chunk.phoneme_timestamps is not None:
+                if chunk.phoneme_timestamps is not None:
+                    phonemes.extend(chunk.phoneme_timestamps.phonemes)
+                    phoneme_start.extend(chunk.phoneme_timestamps.start)
+                    phoneme_end.extend(chunk.phoneme_timestamps.end)
 
         return WebSocketTtsOutput(
             audio=b"".join(chunks),  # type: ignore
@@ -391,6 +423,15 @@ class TtsWebsocket:
                 if add_timestamps
                 else None
             ),
+            phoneme_timestamps=(
+                PhonemeTimestamps(
+                    phonemes=phonemes,
+                    start=phoneme_start,
+                    end=phoneme_end,
+                )
+                if add_phoneme_timestamps
+                else None
+            ),
         )
 
     def _websocket_generator(self, request_body: Dict[str, Any]):

cartesia/tts/client.py

@@ -67,7 +67,7 @@ class TtsClient:
             api_key="YOUR_API_KEY",
         )
         client.tts.bytes(
-            model_id="sonic",
+            model_id="sonic-2",
             transcript="Hello, world!",
             voice={"mode": "id", "id": "694f9389-aac1-45b6-b726-9d9369183238"},
             language="en",
@@ -152,7 +152,7 @@ class TtsClient:
             api_key="YOUR_API_KEY",
         )
         response = client.tts.sse(
-            model_id="sonic",
+            model_id="sonic-2",
             transcript="Hello, world!",
             voice={"mode": "id", "id": "694f9389-aac1-45b6-b726-9d9369183238"},
             language="en",
@@ -258,7 +258,7 @@ class AsyncTtsClient:
 
         async def main() -> None:
             await client.tts.bytes(
-                model_id="sonic",
+                model_id="sonic-2",
                 transcript="Hello, world!",
                 voice={"mode": "id", "id": "694f9389-aac1-45b6-b726-9d9369183238"},
                 language="en",
@@ -351,7 +351,7 @@ class AsyncTtsClient:
 
         async def main() -> None:
             response = await client.tts.sse(
-                model_id="sonic",
+                model_id="sonic-2",
                 transcript="Hello, world!",
                 voice={"mode": "id", "id": "694f9389-aac1-45b6-b726-9d9369183238"},
                 language="en",

cartesia/tts/requests/generation_request.py

@@ -51,3 +51,8 @@ class GenerationRequestParams(typing_extensions.TypedDict):
     """
     Whether to return phoneme-level timestamps.
     """
+
+    use_original_timestamps: typing_extensions.NotRequired[bool]
+    """
+    Whether to use the original transcript for timestamps.
+    """

cartesia/tts/requests/web_socket_chunk_response.py

@@ -1,8 +1,11 @@
 # This file was auto-generated by Fern from our API Definition.
 
 from .web_socket_base_response import WebSocketBaseResponseParams
+import typing_extensions
+from ..types.flush_id import FlushId
 
 
 class WebSocketChunkResponseParams(WebSocketBaseResponseParams):
     data: str
     step_time: float
+    flush_id: typing_extensions.NotRequired[FlushId]

cartesia/tts/requests/web_socket_response.py

@@ -4,8 +4,8 @@ from __future__ import annotations
 import typing_extensions
 import typing
 import typing_extensions
-from ..types.context_id import ContextId
 from ..types.flush_id import FlushId
+from ..types.context_id import ContextId
 from .word_timestamps import WordTimestampsParams
 from .phoneme_timestamps import PhonemeTimestampsParams
 
@@ -14,6 +14,7 @@ class WebSocketResponse_ChunkParams(typing_extensions.TypedDict):
     type: typing.Literal["chunk"]
     data: str
     step_time: float
+    flush_id: typing_extensions.NotRequired[FlushId]
     context_id: typing_extensions.NotRequired[ContextId]
     status_code: int
     done: bool

cartesia/tts/requests/web_socket_tts_request.py

@@ -19,6 +19,7 @@ class WebSocketTtsRequestParams(typing_extensions.TypedDict):
     duration: typing_extensions.NotRequired[int]
     language: typing_extensions.NotRequired[str]
     add_timestamps: typing_extensions.NotRequired[bool]
+    use_original_timestamps: typing_extensions.NotRequired[bool]
     add_phoneme_timestamps: typing_extensions.NotRequired[bool]
     continue_: typing_extensions.NotRequired[typing_extensions.Annotated[bool, FieldMetadata(alias="continue")]]
     context_id: typing_extensions.NotRequired[str]

cartesia/tts/types/emotion.py

@@ -6,22 +6,27 @@ Emotion = typing.Union[
     typing.Literal[
         "anger:lowest",
         "anger:low",
+        "anger",
         "anger:high",
         "anger:highest",
         "positivity:lowest",
         "positivity:low",
+        "positivity",
         "positivity:high",
         "positivity:highest",
         "surprise:lowest",
         "surprise:low",
+        "surprise",
         "surprise:high",
         "surprise:highest",
         "sadness:lowest",
         "sadness:low",
+        "sadness",
         "sadness:high",
         "sadness:highest",
         "curiosity:lowest",
         "curiosity:low",
+        "curiosity",
         "curiosity:high",
         "curiosity:highest",
     ],

cartesia/tts/types/generation_request.py

@@ -56,6 +56,11 @@ class GenerationRequest(UniversalBaseModel):
     Whether to return phoneme-level timestamps.
     """
 
+    use_original_timestamps: typing.Optional[bool] = pydantic.Field(default=None)
+    """
+    Whether to use the original transcript for timestamps.
+    """
+
     if IS_PYDANTIC_V2:
         model_config: typing.ClassVar[pydantic.ConfigDict] = pydantic.ConfigDict(extra="allow", frozen=True)  # type: ignore # Pydantic v2
     else:

cartesia/tts/types/web_socket_chunk_response.py

@@ -1,14 +1,16 @@
 # This file was auto-generated by Fern from our API Definition.
 
 from .web_socket_base_response import WebSocketBaseResponse
-from ...core.pydantic_utilities import IS_PYDANTIC_V2
 import typing
+from .flush_id import FlushId
+from ...core.pydantic_utilities import IS_PYDANTIC_V2
 import pydantic
 
 
 class WebSocketChunkResponse(WebSocketBaseResponse):
     data: str
     step_time: float
+    flush_id: typing.Optional[FlushId] = None
 
     if IS_PYDANTIC_V2:
         model_config: typing.ClassVar[pydantic.ConfigDict] = pydantic.ConfigDict(extra="allow", frozen=True)  # type: ignore # Pydantic v2

cartesia/tts/types/web_socket_response.py

@@ -3,10 +3,10 @@
 from __future__ import annotations
 from ...core.pydantic_utilities import UniversalBaseModel
 import typing
+from .flush_id import FlushId
 from .context_id import ContextId
 from ...core.pydantic_utilities import IS_PYDANTIC_V2
 import pydantic
-from .flush_id import FlushId
 from .word_timestamps import WordTimestamps
 from .phoneme_timestamps import PhonemeTimestamps
 
@@ -15,6 +15,7 @@ class WebSocketResponse_Chunk(UniversalBaseModel):
     type: typing.Literal["chunk"] = "chunk"
     data: str
     step_time: float
+    flush_id: typing.Optional[FlushId] = None
     context_id: typing.Optional[ContextId] = None
     status_code: int
     done: bool

cartesia/tts/types/web_socket_tts_output.py

@@ -7,11 +7,13 @@ import pydantic
 from ...core.pydantic_utilities import IS_PYDANTIC_V2, UniversalBaseModel
 from .context_id import ContextId
 from .flush_id import FlushId
+from .phoneme_timestamps import PhonemeTimestamps
 from .word_timestamps import WordTimestamps
 
 
 class WebSocketTtsOutput(UniversalBaseModel):
     word_timestamps: typing.Optional[WordTimestamps] = None
+    phoneme_timestamps: typing.Optional[PhonemeTimestamps] = None
     audio: typing.Optional[bytes] = None
     context_id: typing.Optional[ContextId] = None
     flush_id: typing.Optional[FlushId] = None

cartesia/tts/types/web_socket_tts_request.py

@@ -22,6 +22,7 @@ class WebSocketTtsRequest(UniversalBaseModel):
     duration: typing.Optional[int] = None
     language: typing.Optional[str] = None
     add_timestamps: typing.Optional[bool] = None
+    use_original_timestamps: typing.Optional[bool] = None
     add_phoneme_timestamps: typing.Optional[bool] = None
     continue_: typing_extensions.Annotated[typing.Optional[bool], FieldMetadata(alias="continue")] = None
     context_id: typing.Optional[str] = None