cartesia 1.3.0__py3-none-any.whl → 1.4.0__py3-none-any.whl

cartesia/_types.py

@@ -36,7 +36,6 @@ class VoiceMetadata(TypedDict):
     user_id: str
     created_at: str
     language: str
-    base_voice_id: Optional[str] = None
 
 
 class VoiceControls(TypedDict):
@@ -62,6 +61,7 @@ class OutputFormat(TypedDict):
     container: str
     encoding: str
     sample_rate: int
+    bit_rate: Optional[int] = None
 
 
 class EventType:

cartesia/_websocket.py

@@ -121,7 +121,7 @@ class _TTSContext:
                         raise RuntimeError(f"Error generating audio:\n{response['error']}")
                     if response["done"]:
                         break
-                    if response["data"]:
+                    if "data" in response and response["data"]:
                         yield self._websocket._convert_response(
                             response=response, include_context_id=True
                         )
@@ -138,7 +138,7 @@ class _TTSContext:
                             raise RuntimeError(f"Error generating audio:\n{response['error']}")
                         if response["done"]:
                             break
-                        if response["data"]:
+                        if "data" in response and response["data"]:
                             yield self._websocket._convert_response(
                                 response=response, include_context_id=True
                             )

cartesia/async_tts.py

@@ -1,4 +1,4 @@
-from typing import Iterator, List, Optional
+from typing import Iterator, List, Optional, Tuple
 
 import httpx
 from cartesia._async_sse import _AsyncSSE
@@ -61,3 +61,116 @@ class AsyncTTS(TTS):
             raise ValueError(f"Failed to generate audio. Error: {response.text}")
 
         return response.content
+
+    async def infill(
+        self,
+        *,
+        model_id: str,
+        language: str,
+        transcript: str,
+        voice_id: str,
+        output_format: OutputFormat,
+        left_audio_path: Optional[str] = None,
+        right_audio_path: Optional[str] = None,
+        experimental_voice_controls: Optional[VoiceControls] = None,
+    ) -> 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")
+
+        headers = self.headers.copy()
+        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[encoding]": output_format["encoding"],
+                "output_format[sample_rate]": output_format["sample_rate"],
+            }
+
+            # Add bit_rate for mp3 container
+            if "bit_rate" in output_format:
+                data["output_format[bit_rate]"] = output_format["bit_rate"]
+
+            # Add voice controls if specified
+            if experimental_voice_controls:
+                if "speed" in experimental_voice_controls:
+                    data["voice[__experimental_controls][speed]"] = experimental_voice_controls[
+                        "speed"
+                    ]
+                if "emotion" in experimental_voice_controls:
+                    # Pass emotions as a list instead of individual values
+                    data["voice[__experimental_controls][emotion][]"] = experimental_voice_controls[
+                        "emotion"
+                    ]
+
+            async with httpx.AsyncClient() as client:
+                response = await client.post(
+                    f"{self._http_url()}/infill/bytes",
+                    headers=headers,
+                    timeout=self.timeout,
+                    files=files,
+                    data=data,
+                )
+
+            if not response.is_success:
+                raise ValueError(
+                    f"Failed to infill audio. Status Code: {response.status_code}\n"
+                    f"Error: {response.text}"
+                )
+
+            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 = self._concat_audio_segments(
+                left_audio, infill_audio, right_audio, format=format
+            )
+            return infill_audio, total_audio
+
+        finally:
+            if left_audio_file:
+                left_audio_file.close()
+            if right_audio_file:
+                right_audio_file.close()

cartesia/tts.py

@@ -1,6 +1,9 @@
-from typing import Iterator, List, Optional
+import json
+from typing import Iterator, List, Optional, Tuple
 
 import httpx
+import io
+from pydub import AudioSegment
 
 from cartesia._sse import _SSE
 from cartesia._types import (
@@ -135,3 +138,155 @@ class TTS(Resource):
             ValueError: If neither or both voice_id and voice_embedding are specified.
         """
         return _validate_and_construct_voice(voice_id, voice_embedding, experimental_voice_controls)
+
+    def infill(
+        self,
+        *,
+        model_id: str,
+        language: str,
+        transcript: str,
+        voice_id: str,
+        output_format: OutputFormat,
+        left_audio_path: Optional[str] = None,
+        right_audio_path: Optional[str] = None,
+        experimental_voice_controls: Optional[VoiceControls] = None,
+    ) -> 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")
+
+        headers = self.headers.copy()
+        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[encoding]": output_format["encoding"],
+                "output_format[sample_rate]": output_format["sample_rate"],
+            }
+
+            # Add bit_rate for mp3 container
+            if "bit_rate" in output_format:
+                data["output_format[bit_rate]"] = output_format["bit_rate"]
+
+            # Add voice controls if specified
+            if experimental_voice_controls:
+                if "speed" in experimental_voice_controls:
+                    data["voice[__experimental_controls][speed]"] = experimental_voice_controls[
+                        "speed"
+                    ]
+                if "emotion" in experimental_voice_controls:
+                    # Pass emotions as a list instead of individual values
+                    data["voice[__experimental_controls][emotion][]"] = experimental_voice_controls[
+                        "emotion"
+                    ]
+
+            response = httpx.post(
+                f"{self._http_url()}/infill/bytes",
+                headers=headers,
+                timeout=self.timeout,
+                files=files,
+                data=data,
+            )
+
+            if not response.is_success:
+                raise ValueError(
+                    f"Failed to infill audio. Status Code: {response.status_code}\n"
+                    f"Error: {response.text}"
+                )
+
+            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 = self._concat_audio_segments(
+                left_audio, infill_audio, right_audio, format=format
+            )
+            return infill_audio, total_audio
+
+        finally:
+            if left_audio_file:
+                left_audio_file.close()
+            if right_audio_file:
+                right_audio_file.close()
+
+    @staticmethod
+    def _concat_audio_segments(
+        left_audio: Optional[bytes],
+        infill_audio: bytes,
+        right_audio: Optional[bytes],
+        format: str = "wav",
+    ) -> bytes:
+        """Helper method to concatenate three audio segments while preserving audio format and headers.
+
+        Args:
+            left_audio: The audio segment that comes before the infill
+            infill_audio: The generated infill audio segment
+            right_audio: The audio segment that comes after the infill
+            format: The audio format (e.g., 'wav', 'mp3'). Defaults to 'wav'
+
+        Returns:
+            bytes: The concatenated audio as bytes
+
+        Raises:
+            ValueError: If the audio segments cannot be loaded or concatenated
+        """
+        try:
+            # Convert bytes to AudioSegment objects
+            combined = AudioSegment.empty()
+            if left_audio:
+                combined += AudioSegment.from_file(io.BytesIO(left_audio), format=format)
+
+            combined += AudioSegment.from_file(io.BytesIO(infill_audio), format=format)
+
+            if right_audio:
+                combined += AudioSegment.from_file(io.BytesIO(right_audio), format=format)
+
+            # Export to bytes
+            output = io.BytesIO()
+            combined.export(output, format=format)
+            return output.getvalue()
+
+        except Exception as e:
+            raise ValueError(f"Failed to concatenate audio segments: {str(e)}")

cartesia/version.py

@@ -1 +1 @@
-__version__ = "1.3.0"
+__version__ = "1.4.0"

cartesia/voices.py

@@ -52,8 +52,7 @@ class Voices(Resource):
 
         if not response.is_success:
             raise ValueError(
-                f"Failed to get voice. Status Code: {response.status_code}\n"
-                f"Error: {response.text}"
+                f"Failed to get voice. Status Code: {response.status_code}\nError: {response.text}"
             )
 
         return response.json()
@@ -123,7 +122,6 @@ class Voices(Resource):
         name: str,
         description: str,
         embedding: List[float],
-        base_voice_id: Optional[str] = None,
         language: str = "en",
     ) -> VoiceMetadata:
         """Create a new voice.
@@ -132,7 +130,6 @@ class Voices(Resource):
             name: The name of the voice.
             description: The description of the voice.
             embedding: The embedding of the voice. This should be generated with :meth:`clone`.
-            base_voice_id: The ID of the base voice. This should be a valid voice ID if specified.
 
         Returns:
             A dictionary containing the voice metadata.
@@ -144,7 +141,6 @@ class Voices(Resource):
                 "name": name,
                 "description": description,
                 "embedding": embedding,
-                "base_voice_id": base_voice_id,
                 "language": language,
             },
             timeout=self.timeout,

{cartesia-1.3.0.dist-info → cartesia-1.4.0.dist-info}/METADATA

@@ -1,12 +1,14 @@
-Metadata-Version: 2.3
+Metadata-Version: 2.4
 Name: cartesia
-Version: 1.3.0
+Version: 1.4.0
 Summary: The official Python library for the Cartesia API.
+License-File: LICENSE.md
 Requires-Python: >=3.9
 Requires-Dist: aiohttp>=3.10.10
-Requires-Dist: httpx>=0.27.2
+Requires-Dist: httpx>=0.27.0
 Requires-Dist: iterators>=0.2.0
-Requires-Dist: requests>=2.32.3
+Requires-Dist: pydub>=0.25.1
+Requires-Dist: requests>=2.31.0
 Requires-Dist: websockets>=10.4
 Description-Content-Type: text/markdown
 

{cartesia-1.3.0.dist-info → cartesia-1.4.0.dist-info}/RECORD

@@ -4,20 +4,20 @@ cartesia/_async_websocket.py,sha256=y9YL9fU8eLENZZECJUwRBVTfEx4ZMl96Y5zHaRY2BiI,
 cartesia/_constants.py,sha256=khGNVpiQVDmv1oZU7pKTd9C1AHjiaM8zQ2He9d5zI_c,435
 cartesia/_logger.py,sha256=vU7QiGSy_AJuJFmClUocqIJ-Ltku_8C24ZU8L6fLJR0,53
 cartesia/_sse.py,sha256=CugabGUAUM-N2BruxNFxDB20HyxDlRdbN-J_yAzvBMY,5667
-cartesia/_types.py,sha256=gixQbKbX-H8xbD7jxHmc02KXLyjEaup19lh_57_YBl8,2570
-cartesia/_websocket.py,sha256=nRCq9xB0T9yYHoLqtn0GsJmcap-OAlJdSIrzTl40qMI,14875
+cartesia/_types.py,sha256=p0OzSzH174WrG8LRyu_MvNXZPhhTfLArSSDpcYY4xa0,2565
+cartesia/_websocket.py,sha256=7gDLcfMoIwmKj07iLk5UZ4ypxlv-3UmMd3VFjVn1QaE,14921
 cartesia/async_client.py,sha256=y_K_Yuv0weA4k9ZYD0M9bNM3x3frsq07tqkg7R9h0-o,2714
-cartesia/async_tts.py,sha256=IbWVRKklNClXASR6ylHaukcMRR304LUguqc4yMopbDU,2076
+cartesia/async_tts.py,sha256=CgbrLk7tc0NKSBC8zZH5I4CpWHpOgkypo0D2hyg5LLE,6466
 cartesia/client.py,sha256=OS1ORUSlR8Jg-em1imeTAFfwkC85AQFnw8PYtTdUuC8,2364
 cartesia/resource.py,sha256=wpnB3IPcTdxYSp0vxSkpntp4NSvqvnwUWF-0ZpgWV9o,1585
-cartesia/tts.py,sha256=kWvqce9K3gZ4QrWD-ciYdK29n49SNkxhd2A7ueTOwMY,4878
-cartesia/version.py,sha256=F5mW07pSyGrqDNY2Ehr-UpDzpBtN-FsYU0QGZWf6PJE,22
-cartesia/voices.py,sha256=bDYbs0KoikAROJlmbnLdo4TrW0YwzjMvp70uKG6Alp0,7180
+cartesia/tts.py,sha256=WV8OduM87ciM1ht60Fi9Fh4gunX2Xew3K96ELCzpP-8,10702
+cartesia/version.py,sha256=8UhoYEXHs1Oai7BW_ExBmuwWnRI-yMG_u1fQAXMizHQ,22
+cartesia/voices.py,sha256=DLO_GJYDRhzFbqVIqzGOP1m1Ylzq7tVm6VHrknekFCk,6968
 cartesia/utils/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 cartesia/utils/deprecated.py,sha256=2cXvGtrxhPeUZA5LWy2n_U5OFLDv7SHeFtzqhjSJGyk,1674
 cartesia/utils/retry.py,sha256=O6fyVWpH9Su8c0Fwupl57xMt6JrwJ52txBwP3faUL7k,3339
 cartesia/utils/tts.py,sha256=TbvBZqHR6LxPim6s5RyGiURi4hIfqWt3KUk5QYOOhfc,2177
-cartesia-1.3.0.dist-info/METADATA,sha256=eedG5B4V6MxvDuPUMdYwp6UHrX6yQ6dJTMRRZxq1-UA,20976
-cartesia-1.3.0.dist-info/WHEEL,sha256=C2FUgwZgiLbznR-k0b_5k3Ai_1aASOXDss3lzCUsUug,87
-cartesia-1.3.0.dist-info/licenses/LICENSE.md,sha256=PT2YG5wEtEX1TNDn5sXkUXqbn-neyr7cZenTxd40ql4,1074
-cartesia-1.3.0.dist-info/RECORD,,
+cartesia-1.4.0.dist-info/METADATA,sha256=LLv3iE6dKcAeZuSb6bDAc0GSVICUOS8szaM4n1F-Mww,21030
+cartesia-1.4.0.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
+cartesia-1.4.0.dist-info/licenses/LICENSE.md,sha256=PT2YG5wEtEX1TNDn5sXkUXqbn-neyr7cZenTxd40ql4,1074
+cartesia-1.4.0.dist-info/RECORD,,

{cartesia-1.3.0.dist-info → cartesia-1.4.0.dist-info}/WHEEL

@@ -1,4 +1,4 @@
 Wheel-Version: 1.0
-Generator: hatchling 1.26.3
+Generator: hatchling 1.27.0
 Root-Is-Purelib: true
 Tag: py3-none-any