cartesia 1.4.0__py3-none-any.whl → 2.0.0__py3-none-any.whl

cartesia/tts/requests/tts_request_id_specifier.py

@@ -0,0 +1,16 @@
+# This file was auto-generated by Fern from our API Definition.
+
+import typing_extensions
+import typing
+from ...voices.types.voice_id import VoiceId
+import typing_extensions
+from .controls import ControlsParams
+from ...core.serialization import FieldMetadata
+
+
+class TtsRequestIdSpecifierParams(typing_extensions.TypedDict):
+    mode: typing.Literal["id"]
+    id: VoiceId
+    experimental_controls: typing_extensions.NotRequired[
+        typing_extensions.Annotated[ControlsParams, FieldMetadata(alias="__experimental_controls")]
+    ]

cartesia/tts/requests/tts_request_voice_specifier.py

@@ -0,0 +1,7 @@
+# This file was auto-generated by Fern from our API Definition.
+
+import typing
+from .tts_request_id_specifier import TtsRequestIdSpecifierParams
+from .tts_request_embedding_specifier import TtsRequestEmbeddingSpecifierParams
+
+TtsRequestVoiceSpecifierParams = typing.Union[TtsRequestIdSpecifierParams, TtsRequestEmbeddingSpecifierParams]

cartesia/tts/requests/wav_output_format.py

@@ -0,0 +1,7 @@
+# This file was auto-generated by Fern from our API Definition.
+
+from .raw_output_format import RawOutputFormatParams
+
+
+class WavOutputFormatParams(RawOutputFormatParams):
+    pass

cartesia/tts/requests/web_socket_base_response.py

@@ -0,0 +1,11 @@
+# This file was auto-generated by Fern from our API Definition.
+
+import typing_extensions
+import typing_extensions
+from ..types.context_id import ContextId
+
+
+class WebSocketBaseResponseParams(typing_extensions.TypedDict):
+    context_id: typing_extensions.NotRequired[ContextId]
+    status_code: int
+    done: bool

cartesia/tts/requests/web_socket_chunk_response.py

@@ -0,0 +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_done_response.py

@@ -0,0 +1,7 @@
+# This file was auto-generated by Fern from our API Definition.
+
+from .web_socket_base_response import WebSocketBaseResponseParams
+
+
+class WebSocketDoneResponseParams(WebSocketBaseResponseParams):
+    pass

cartesia/tts/requests/web_socket_error_response.py

@@ -0,0 +1,7 @@
+# This file was auto-generated by Fern from our API Definition.
+
+from .web_socket_base_response import WebSocketBaseResponseParams
+
+
+class WebSocketErrorResponseParams(WebSocketBaseResponseParams):
+    error: str

cartesia/tts/requests/web_socket_flush_done_response.py

@@ -0,0 +1,9 @@
+# This file was auto-generated by Fern from our API Definition.
+
+from .web_socket_base_response import WebSocketBaseResponseParams
+from ..types.flush_id import FlushId
+
+
+class WebSocketFlushDoneResponseParams(WebSocketBaseResponseParams):
+    flush_id: FlushId
+    flush_done: bool

cartesia/tts/requests/web_socket_phoneme_timestamps_response.py

@@ -0,0 +1,9 @@
+# This file was auto-generated by Fern from our API Definition.
+
+from .web_socket_base_response import WebSocketBaseResponseParams
+import typing_extensions
+from .phoneme_timestamps import PhonemeTimestampsParams
+
+
+class WebSocketPhonemeTimestampsResponseParams(WebSocketBaseResponseParams):
+    phoneme_timestamps: typing_extensions.NotRequired[PhonemeTimestampsParams]

cartesia/tts/requests/web_socket_raw_output_format.py

@@ -0,0 +1,11 @@
+# This file was auto-generated by Fern from our API Definition.
+
+import typing_extensions
+import typing
+from ..types.raw_encoding import RawEncoding
+
+
+class WebSocketRawOutputFormatParams(typing_extensions.TypedDict):
+    container: typing.Literal["raw"]
+    encoding: RawEncoding
+    sample_rate: int

cartesia/tts/requests/web_socket_request.py

@@ -0,0 +1,7 @@
+# This file was auto-generated by Fern from our API Definition.
+
+import typing
+from .generation_request import GenerationRequestParams
+from .cancel_context_request import CancelContextRequestParams
+
+WebSocketRequestParams = typing.Union[GenerationRequestParams, CancelContextRequestParams]

cartesia/tts/requests/web_socket_response.py

@@ -0,0 +1,70 @@
+# This file was auto-generated by Fern from our API Definition.
+
+from __future__ import annotations
+import typing_extensions
+import typing
+import typing_extensions
+from ..types.flush_id import FlushId
+from ..types.context_id import ContextId
+from .word_timestamps import WordTimestampsParams
+from .phoneme_timestamps import PhonemeTimestampsParams
+
+
+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
+
+
+class WebSocketResponse_FlushDoneParams(typing_extensions.TypedDict):
+    type: typing.Literal["flush_done"]
+    flush_id: FlushId
+    flush_done: bool
+    context_id: typing_extensions.NotRequired[ContextId]
+    status_code: int
+    done: bool
+
+
+class WebSocketResponse_DoneParams(typing_extensions.TypedDict):
+    type: typing.Literal["done"]
+    context_id: typing_extensions.NotRequired[ContextId]
+    status_code: int
+    done: bool
+
+
+class WebSocketResponse_TimestampsParams(typing_extensions.TypedDict):
+    type: typing.Literal["timestamps"]
+    word_timestamps: typing_extensions.NotRequired[WordTimestampsParams]
+    context_id: typing_extensions.NotRequired[ContextId]
+    status_code: int
+    done: bool
+
+
+class WebSocketResponse_ErrorParams(typing_extensions.TypedDict):
+    type: typing.Literal["error"]
+    error: str
+    context_id: typing_extensions.NotRequired[ContextId]
+    status_code: int
+    done: bool
+
+
+class WebSocketResponse_PhonemeTimestampsParams(typing_extensions.TypedDict):
+    type: typing.Literal["phoneme_timestamps"]
+    phoneme_timestamps: typing_extensions.NotRequired[PhonemeTimestampsParams]
+    context_id: typing_extensions.NotRequired[ContextId]
+    status_code: int
+    done: bool
+
+
+WebSocketResponseParams = typing.Union[
+    WebSocketResponse_ChunkParams,
+    WebSocketResponse_FlushDoneParams,
+    WebSocketResponse_DoneParams,
+    WebSocketResponse_TimestampsParams,
+    WebSocketResponse_ErrorParams,
+    WebSocketResponse_PhonemeTimestampsParams,
+]

cartesia/tts/requests/web_socket_stream_options.py

@@ -0,0 +1,8 @@
+# This file was auto-generated by Fern from our API Definition.
+
+import typing_extensions
+import typing_extensions
+
+
+class WebSocketStreamOptionsParams(typing_extensions.TypedDict):
+    timeout: typing_extensions.NotRequired[float]

cartesia/tts/requests/web_socket_timestamps_response.py

@@ -0,0 +1,9 @@
+# This file was auto-generated by Fern from our API Definition.
+
+from .web_socket_base_response import WebSocketBaseResponseParams
+import typing_extensions
+from .word_timestamps import WordTimestampsParams
+
+
+class WebSocketTimestampsResponseParams(WebSocketBaseResponseParams):
+    word_timestamps: typing_extensions.NotRequired[WordTimestampsParams]

cartesia/tts/requests/web_socket_tts_output.py

@@ -0,0 +1,18 @@
+# This file was auto-generated by Fern from our API Definition.
+
+import typing_extensions
+import typing_extensions
+from .word_timestamps import WordTimestampsParams
+from .phoneme_timestamps import PhonemeTimestampsParams
+import typing
+from ..types.context_id import ContextId
+from ..types.flush_id import FlushId
+
+
+class WebSocketTtsOutputParams(typing_extensions.TypedDict):
+    word_timestamps: typing_extensions.NotRequired[WordTimestampsParams]
+    phoneme_timestamps: typing_extensions.NotRequired[PhonemeTimestampsParams]
+    audio: typing_extensions.NotRequired[typing.Optional[typing.Any]]
+    context_id: typing_extensions.NotRequired[ContextId]
+    flush_id: typing_extensions.NotRequired[FlushId]
+    flush_done: typing_extensions.NotRequired[bool]

cartesia/tts/requests/web_socket_tts_request.py

@@ -0,0 +1,25 @@
+# This file was auto-generated by Fern from our API Definition.
+
+import typing_extensions
+import typing_extensions
+from .output_format import OutputFormatParams
+from .tts_request_voice_specifier import TtsRequestVoiceSpecifierParams
+from ...core.serialization import FieldMetadata
+
+
+class WebSocketTtsRequestParams(typing_extensions.TypedDict):
+    model_id: str
+    """
+    The ID of the model to use for the generation. See [Models](/build-with-sonic/models) for available models.
+    """
+
+    output_format: typing_extensions.NotRequired[OutputFormatParams]
+    transcript: typing_extensions.NotRequired[str]
+    voice: TtsRequestVoiceSpecifierParams
+    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/requests/word_timestamps.py

@@ -0,0 +1,10 @@
+# This file was auto-generated by Fern from our API Definition.
+
+import typing_extensions
+import typing
+
+
+class WordTimestampsParams(typing_extensions.TypedDict):
+    words: typing.Sequence[str]
+    start: typing.Sequence[float]
+    end: typing.Sequence[float]

cartesia/tts/socket_client.py

@@ -0,0 +1,302 @@
+import io
+import typing
+from json.decoder import JSONDecodeError
+
+from pydub import AudioSegment  # type: ignore
+
+from ..core.api_error import ApiError
+from ._async_websocket import AsyncTtsWebsocket
+from ._websocket import TtsWebsocket
+from .client import AsyncTtsClient, TtsClient
+from .requests import TtsRequestVoiceSpecifierParams
+from .requests.output_format import OutputFormatParams
+from .utils.tts import concat_audio_segments, get_output_format
+
+
+class TtsClientWithWebsocket(TtsClient):
+    """
+    Extension of TtsClient that supports a synchronous WebSocket TTS connection.
+    """
+
+    def __init__(self, *, client_wrapper):
+        super().__init__(client_wrapper=client_wrapper)
+
+    def get_output_format(self, output_format_name: str) -> OutputFormatParams:
+        return get_output_format(output_format_name)
+
+    def _ws_url(self):
+        base_url = self._client_wrapper.get_base_url()
+        if base_url.startswith("ws://") or base_url.startswith("wss://"):
+            return base_url
+        else:
+            prefix = "ws" if "localhost" in base_url else "wss"
+            base_url_without_protocol = base_url.split("://")[-1]
+            return f"{prefix}://{base_url_without_protocol}"
+
+    def infill(
+        self,
+        *,
+        model_id: str,
+        language: str,
+        transcript: str,
+        voice: TtsRequestVoiceSpecifierParams,
+        output_format: OutputFormatParams,
+        left_audio_path: typing.Optional[str] = None,
+        right_audio_path: typing.Optional[str] = None,
+    ) -> typing.Tuple[bytes, bytes]:
+        """Generate infill audio between two existing audio segments.
+
+        Args:
+            model_id: The ID of the model to use for generating audio
+            language: The language of the transcript
+            transcript: The text to synthesize
+            voice: The voice to use for generating audio
+            output_format: The desired audio output format
+            left_audio_path: Path to the audio file that comes before the infill
+            right_audio_path: Path to the audio file that comes after the infill
+
+        Returns:
+            A tuple containing:
+            - The generated infill audio (bytes)
+            - The complete concatenated audio (bytes)
+        """
+        if not left_audio_path and not right_audio_path:
+            raise ValueError(
+                "Must specify at least one of left_audio_path or right_audio_path"
+            )
+
+        if voice["mode"] != "id":
+            raise ValueError("Infill is only supported for id-based voice specifiers")
+
+        if output_format["container"] == "raw":
+            raise ValueError(
+                "Raw format is not supported for infill. Use wav or mp3 format instead."
+            )
+
+        headers = self._client_wrapper.get_headers()
+        headers.pop("Content-Type", None)
+
+        left_audio_file = None
+        right_audio_file = None
+        try:
+            files = {}
+            if left_audio_path:
+                left_audio_file = open(left_audio_path, "rb")
+                files["left_audio"] = left_audio_file
+            if right_audio_path:
+                right_audio_file = open(right_audio_path, "rb")
+                files["right_audio"] = right_audio_file
+
+            # Construct form data with output_format fields directly
+            data = {
+                "model_id": model_id,
+                "language": language,
+                "transcript": transcript,
+                "voice_id": voice["id"],
+                "output_format[container]": output_format["container"],
+                "output_format[sample_rate]": output_format["sample_rate"],
+            }
+
+            # Add bit_rate for mp3 container
+            if "bit_rate" in output_format and output_format["bit_rate"] is not None:
+                data["output_format[bit_rate]"] = output_format["bit_rate"]
+            if (
+                output_format["container"] != "mp3"
+                and "encoding" in output_format
+                and output_format["encoding"] is not None
+            ):
+                data["output_format[encoding]"] = output_format["encoding"]
+
+            _response = self._client_wrapper.httpx_client.request(
+                "infill/bytes",
+                method="POST",
+                files=files,  # type: ignore
+                data=data,
+                headers=headers,
+            )
+            try:
+                if 200 <= _response.status_code < 300:
+                    if left_audio_file:
+                        left_audio_file.seek(0)
+                        left_audio = left_audio_file.read()
+                    else:
+                        left_audio = None
+
+                    if right_audio_file:
+                        right_audio_file.seek(0)
+                        right_audio = right_audio_file.read()
+                    else:
+                        right_audio = None
+
+                    infill_audio = _response.content
+                    format = output_format["container"].lower()
+                    total_audio = concat_audio_segments(
+                        left_audio, infill_audio, right_audio, format=format
+                    )
+                    return infill_audio, total_audio
+
+                _response_json = _response.json()
+            except JSONDecodeError:
+                raise ApiError(status_code=_response.status_code, body=_response.text)
+            raise ApiError(status_code=_response.status_code, body=_response_json)
+
+        finally:
+            if left_audio_file:
+                left_audio_file.close()
+            if right_audio_file:
+                right_audio_file.close()
+
+    def websocket(self):
+        client_headers = self._client_wrapper.get_headers()
+        ws = TtsWebsocket(
+            ws_url=self._ws_url(),
+            cartesia_version=client_headers["Cartesia-Version"],
+            api_key=client_headers["X-API-Key"],
+        )
+        ws.connect()
+        return ws
+
+
+class AsyncTtsClientWithWebsocket(AsyncTtsClient):
+    """
+    Extension of AsyncTtsClient that supports an asynchronous WebSocket TTS connection.
+    """
+
+    def __init__(self, *, client_wrapper, get_session):
+        super().__init__(client_wrapper=client_wrapper)
+        self._get_session = get_session
+
+    def get_output_format(self, output_format_name: str) -> OutputFormatParams:
+        return get_output_format(output_format_name)
+
+    def _ws_url(self) -> str:
+        base_url = self._client_wrapper.get_base_url()
+        if base_url.startswith("ws://") or base_url.startswith("wss://"):
+            return base_url
+        else:
+            prefix = "ws" if "localhost" in base_url else "wss"
+            base_url_without_protocol = base_url.split("://")[-1]
+            return f"{prefix}://{base_url_without_protocol}"
+
+    async def infill(
+        self,
+        *,
+        model_id: str,
+        language: str,
+        transcript: str,
+        voice: TtsRequestVoiceSpecifierParams,
+        output_format: OutputFormatParams,
+        left_audio_path: typing.Optional[str] = None,
+        right_audio_path: typing.Optional[str] = None,
+    ) -> typing.Tuple[bytes, bytes]:
+        """Generate infill audio between two existing audio segments.
+        Args:
+            model_id: The ID of the model to use for generating audio
+            language: The language of the transcript
+            transcript: The text to synthesize
+            voice_id: The ID of the voice to use for generating audio
+            output_format: The desired audio output format
+            left_audio_path: Path to the audio file that comes before the infill
+            right_audio_path: Path to the audio file that comes after the infill
+            experimental_voice_controls: Optional voice control parameters
+        Returns:
+            A tuple containing:
+            - The generated infill audio (bytes)
+            - The complete concatenated audio (bytes)
+        """
+        if not left_audio_path and not right_audio_path:
+            raise ValueError(
+                "Must specify at least one of left_audio_path or right_audio_path"
+            )
+
+        if voice["mode"] != "id":
+            raise ValueError("Infill is only supported for id-based voice specifiers")
+
+        if output_format["container"] == "raw":
+            raise ValueError(
+                "Raw format is not supported for infill. Use wav or mp3 format instead."
+            )
+
+        headers = self._client_wrapper.get_headers()
+        headers.pop("Content-Type", None)
+
+        left_audio_file = None
+        right_audio_file = None
+        try:
+            files = {}
+            if left_audio_path:
+                left_audio_file = open(left_audio_path, "rb")
+                files["left_audio"] = left_audio_file
+            if right_audio_path:
+                right_audio_file = open(right_audio_path, "rb")
+                files["right_audio"] = right_audio_file
+
+            data = {
+                "model_id": model_id,
+                "language": language,
+                "transcript": transcript,
+                "voice_id": voice["id"],
+                "output_format[container]": output_format["container"],
+                "output_format[sample_rate]": output_format["sample_rate"],
+            }
+
+            if "bit_rate" in output_format and output_format["bit_rate"] is not None:
+                data["output_format[bit_rate]"] = output_format["bit_rate"]
+            if (
+                output_format["container"] != "mp3"
+                and "encoding" in output_format
+                and output_format["encoding"] is not None
+            ):
+                data["output_format[encoding]"] = output_format["encoding"]
+
+            _response = await self._client_wrapper.httpx_client.request(
+                "infill/bytes",
+                method="POST",
+                files=files,  # type: ignore
+                headers=headers,
+                data=data,
+                request_options=None,
+            )
+
+            try:
+                if 200 <= _response.status_code < 300:
+                    if left_audio_file:
+                        left_audio_file.seek(0)
+                        left_audio = left_audio_file.read()
+                    else:
+                        left_audio = None
+
+                    if right_audio_file:
+                        right_audio_file.seek(0)
+                        right_audio = right_audio_file.read()
+                    else:
+                        right_audio = None
+
+                    infill_audio = _response.content
+                    audio_format = output_format["container"].lower()
+                    total_audio = concat_audio_segments(
+                        left_audio, infill_audio, right_audio, format=audio_format
+                    )
+                    return infill_audio, total_audio
+
+                _response_json = _response.json()
+            except JSONDecodeError:
+                raise ApiError(status_code=_response.status_code, body=_response.text)
+            raise ApiError(status_code=_response.status_code, body=_response_json)
+
+        finally:
+            if left_audio_file:
+                left_audio_file.close()
+            if right_audio_file:
+                right_audio_file.close()
+
+    async def websocket(self):
+        client_headers = self._client_wrapper.get_headers()
+        ws = AsyncTtsWebsocket(
+            ws_url=self._ws_url(),
+            cartesia_version=client_headers["Cartesia-Version"],
+            api_key=client_headers["X-API-Key"],
+            get_session=self._get_session,
+        )
+        await ws.connect()
+        return ws

cartesia/tts/types/__init__.py

@@ -0,0 +1,90 @@
+# This file was auto-generated by Fern from our API Definition.
+
+from .cancel_context_request import CancelContextRequest
+from .context_id import ContextId
+from .controls import Controls
+from .emotion import Emotion
+from .flush_id import FlushId
+from .generation_request import GenerationRequest
+from .mp_3_output_format import Mp3OutputFormat
+from .natural_specifier import NaturalSpecifier
+from .numerical_specifier import NumericalSpecifier
+from .output_format import OutputFormat, OutputFormat_Mp3, OutputFormat_Raw, OutputFormat_Wav
+from .phoneme_timestamps import PhonemeTimestamps
+from .raw_encoding import RawEncoding
+from .raw_output_format import RawOutputFormat
+from .speed import Speed
+from .supported_language import SupportedLanguage
+from .tts_request import TtsRequest
+from .tts_request_embedding_specifier import TtsRequestEmbeddingSpecifier
+from .tts_request_id_specifier import TtsRequestIdSpecifier
+from .tts_request_voice_specifier import TtsRequestVoiceSpecifier
+from .wav_output_format import WavOutputFormat
+from .web_socket_base_response import WebSocketBaseResponse
+from .web_socket_chunk_response import WebSocketChunkResponse
+from .web_socket_done_response import WebSocketDoneResponse
+from .web_socket_error_response import WebSocketErrorResponse
+from .web_socket_flush_done_response import WebSocketFlushDoneResponse
+from .web_socket_phoneme_timestamps_response import WebSocketPhonemeTimestampsResponse
+from .web_socket_raw_output_format import WebSocketRawOutputFormat
+from .web_socket_request import WebSocketRequest
+from .web_socket_response import (
+    WebSocketResponse,
+    WebSocketResponse_Chunk,
+    WebSocketResponse_Done,
+    WebSocketResponse_Error,
+    WebSocketResponse_FlushDone,
+    WebSocketResponse_PhonemeTimestamps,
+    WebSocketResponse_Timestamps,
+)
+from .web_socket_stream_options import WebSocketStreamOptions
+from .web_socket_timestamps_response import WebSocketTimestampsResponse
+from .web_socket_tts_output import WebSocketTtsOutput
+from .web_socket_tts_request import WebSocketTtsRequest
+from .word_timestamps import WordTimestamps
+
+__all__ = [
+    "CancelContextRequest",
+    "ContextId",
+    "Controls",
+    "Emotion",
+    "FlushId",
+    "GenerationRequest",
+    "Mp3OutputFormat",
+    "NaturalSpecifier",
+    "NumericalSpecifier",
+    "OutputFormat",
+    "OutputFormat_Mp3",
+    "OutputFormat_Raw",
+    "OutputFormat_Wav",
+    "PhonemeTimestamps",
+    "RawEncoding",
+    "RawOutputFormat",
+    "Speed",
+    "SupportedLanguage",
+    "TtsRequest",
+    "TtsRequestEmbeddingSpecifier",
+    "TtsRequestIdSpecifier",
+    "TtsRequestVoiceSpecifier",
+    "WavOutputFormat",
+    "WebSocketBaseResponse",
+    "WebSocketChunkResponse",
+    "WebSocketDoneResponse",
+    "WebSocketErrorResponse",
+    "WebSocketFlushDoneResponse",
+    "WebSocketPhonemeTimestampsResponse",
+    "WebSocketRawOutputFormat",
+    "WebSocketRequest",
+    "WebSocketResponse",
+    "WebSocketResponse_Chunk",
+    "WebSocketResponse_Done",
+    "WebSocketResponse_Error",
+    "WebSocketResponse_FlushDone",
+    "WebSocketResponse_PhonemeTimestamps",
+    "WebSocketResponse_Timestamps",
+    "WebSocketStreamOptions",
+    "WebSocketTimestampsResponse",
+    "WebSocketTtsOutput",
+    "WebSocketTtsRequest",
+    "WordTimestamps",
+]