sinapsis-speech 0.3.5__tar.gz → 0.4.0__tar.gz

{sinapsis_speech-0.3.5 → sinapsis_speech-0.4.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: sinapsis-speech
-Version: 0.3.5
+Version: 0.4.0
 Summary: Generate speech using various libraries.
 Author-email: SinapsisAI <dev@sinapsis.tech>
 Project-URL: Homepage, https://sinapsis.tech
@@ -18,6 +18,8 @@ Requires-Dist: sinapsis-f5-tts[all]; extra == "all"
 Requires-Dist: sinapsis-kokoro[all]; extra == "all"
 Requires-Dist: sinapsis-speech[gradio-app]; extra == "all"
 Requires-Dist: sinapsis-zonos[all]; extra == "all"
+Requires-Dist: sinapsis-parakeet-tdt[all]; extra == "all"
+Requires-Dist: sinapsis-orpheus-cpp[all]; extra == "all"
 Provides-Extra: gradio-app
 Requires-Dist: sinapsis[webapp]>=0.2.3; extra == "gradio-app"
 Dynamic: license-file
@@ -55,8 +57,10 @@ This repo includes packages for performing speech synthesis using different tool
 
 * <code>sinapsis-elevenlabs</code>
 * <code>sinapsis-f5-tts</code>
-* * <code>sinapsis-kokoro</code>
+* <code>sinapsis-kokoro</code>
 * <code>sinapsis-zonos</code>
+* <code>sinapsis-orpheus-cpp</code>
+* <code>sinapsis-parakeet</code>
 
 Install using your preferred package manager. We strongly recommend using <code>uv</code>. To install <code>uv</code>, refer to the [official documentation](https://docs.astral.sh/uv/getting-started/installation/#installation-methods).
 
@@ -205,6 +209,16 @@ docker compose -f docker/compose_apps.yaml up -d sinapsis-kokoro
 docker compose -f docker/compose_apps.yaml up -d sinapsis-zonos
 ```
 
+- For Orpheus-CPP:
+```bash
+docker compose -f docker/compose_apps.yaml up -d sinapsis-orpheus-tts
+```
+
+- For Parakeet:
+```bash
+docker compose -f docker/compose_apps.yaml up -d sinapsis-parakeet
+```
+
 3. **Check the logs**
 
 - For ElevenLabs:
@@ -224,6 +238,17 @@ docker logs -f sinapsis-kokoro
 ```bash
 docker logs -f sinapsis-zonos
 ```
+
+- For Orpheus-CPP:
+```bash
+docker logs -f sinapsis-orpheus-tts
+```
+
+- For Parakeet:
+```bash
+docker logs -f sinapsis-parakeet
+```
+
 4. **The logs will display the URL to access the webapp, e.g.,:**:
 ```bash
 Running on local URL:  http://127.0.0.1:7860
@@ -240,6 +265,17 @@ docker compose -f docker/compose_apps.yaml down
 
 To run the webapp using the <code>uv</code> package manager, follow these steps:
 
+
+> [!IMPORTANT]
+> If you're using sinapsis-orpheus-cpp, you need to export cuda environment variables:
+
+
+```bash
+export CMAKE_ARGS="-DGGML_CUDA=on"
+export FORCE_CMAKE="1"
+export CUDACXX=$(command -v nvcc)
+```
+
 1. **Sync the virtual environment**:
 
 ```bash

{sinapsis_speech-0.3.5 → sinapsis_speech-0.4.0}/README.md

@@ -31,8 +31,10 @@ This repo includes packages for performing speech synthesis using different tool
 
 * <code>sinapsis-elevenlabs</code>
 * <code>sinapsis-f5-tts</code>
-* * <code>sinapsis-kokoro</code>
+* <code>sinapsis-kokoro</code>
 * <code>sinapsis-zonos</code>
+* <code>sinapsis-orpheus-cpp</code>
+* <code>sinapsis-parakeet</code>
 
 Install using your preferred package manager. We strongly recommend using <code>uv</code>. To install <code>uv</code>, refer to the [official documentation](https://docs.astral.sh/uv/getting-started/installation/#installation-methods).
 
@@ -181,6 +183,16 @@ docker compose -f docker/compose_apps.yaml up -d sinapsis-kokoro
 docker compose -f docker/compose_apps.yaml up -d sinapsis-zonos
 ```
 
+- For Orpheus-CPP:
+```bash
+docker compose -f docker/compose_apps.yaml up -d sinapsis-orpheus-tts
+```
+
+- For Parakeet:
+```bash
+docker compose -f docker/compose_apps.yaml up -d sinapsis-parakeet
+```
+
 3. **Check the logs**
 
 - For ElevenLabs:
@@ -200,6 +212,17 @@ docker logs -f sinapsis-kokoro
 ```bash
 docker logs -f sinapsis-zonos
 ```
+
+- For Orpheus-CPP:
+```bash
+docker logs -f sinapsis-orpheus-tts
+```
+
+- For Parakeet:
+```bash
+docker logs -f sinapsis-parakeet
+```
+
 4. **The logs will display the URL to access the webapp, e.g.,:**:
 ```bash
 Running on local URL:  http://127.0.0.1:7860
@@ -216,6 +239,17 @@ docker compose -f docker/compose_apps.yaml down
 
 To run the webapp using the <code>uv</code> package manager, follow these steps:
 
+
+> [!IMPORTANT]
+> If you're using sinapsis-orpheus-cpp, you need to export cuda environment variables:
+
+
+```bash
+export CMAKE_ARGS="-DGGML_CUDA=on"
+export FORCE_CMAKE="1"
+export CUDACXX=$(command -v nvcc)
+```
+
 1. **Sync the virtual environment**:
 
 ```bash

sinapsis_speech-0.4.0/packages/sinapsis_elevenlabs/src/sinapsis_elevenlabs/helpers/voice_utils.py

@@ -0,0 +1,100 @@
+# -*- coding: utf-8 -*-
+import json
+
+from elevenlabs import Voice, VoiceSettings
+from elevenlabs.client import ElevenLabs
+from sinapsis_core.data_containers.data_packet import TextPacket
+from sinapsis_core.utils.logging_utils import sinapsis_logger
+
+
+def create_voice_settings(settings: VoiceSettings, as_json: bool = False) -> VoiceSettings | None | str:
+    """
+    Creates or updates a `VoiceSettings` object based on the provided settings.
+
+    Args:
+        settings (VoiceSettings | None): An instance of `VoiceSettings` containing the settings to be applied.
+            If `None`, the function returns the default settings.
+        as_json (bool): Whether to return the settings as JSON string.
+
+    Returns:
+        VoiceSettings | None | str: The provided `VoiceSettings` object if `settings` is not `None`. Otherwise,
+            `None` is returned for default settings.
+    """
+    if not settings:
+        return None
+
+    if as_json:
+        return json.dumps(settings.model_dump(exclude_none=True))
+
+    return settings
+
+
+def get_voice_id(client: ElevenLabs, voice: str | Voice | None) -> str:
+    """
+    Resolves the voice ID for a given voice name or ID.
+
+    This function searches through available voices from the ElevenLabs API
+    to match the provided voice name or ID. If the specified voice is not found,
+    it logs the error and returns the first available voice ID as a fallback.
+
+    Args:
+        client (ElevenLabs): The ElevenLabs API client instance.
+        voice (str | Voice | None): The name or ID of the desired voice.
+
+    Returns:
+        str: The resolved voice ID.
+
+    Raises:
+        ValueError: If no voices are available to resolve.
+    """
+    if not voice:
+        return get_default_voice(client).voice_id
+
+    if isinstance(voice, Voice):
+        sinapsis_logger.debug(f"Voice object provided, using voice_id: {voice.voice_id}")
+        return voice.voice_id
+
+    try:
+        voices_response = client.voices.get_all()
+        voices = voices_response.voices
+
+        for v in voices:
+            if voice == v.name or voice == v.voice_id:
+                sinapsis_logger.debug(f"Voice {voice} resolved to ID: {v.voice_id}")
+                return v.voice_id
+
+        sinapsis_logger.error(f"Voice {voice} is not available.")
+        if voices:
+            sinapsis_logger.info(f"Returning default voice ID: {voices[0].voice_id}")
+            return voices[0].voice_id
+
+        raise ValueError("No available voices to resolve. Ensure the client is configured correctly.")
+    except Exception as e:
+        sinapsis_logger.error(f"Error resolving voice ID: {e}")
+        raise
+
+
+def get_default_voice(client: ElevenLabs) -> Voice:
+    """
+    Gets the first available voice as default.
+
+    Args:
+        client (ElevenLabs): The ElevenLabs API client instance.
+
+    Returns:
+        Voice: The default voice object.
+    """
+    try:
+        voices_response = client.voices.get_all()
+        voices = voices_response.voices
+        if voices:
+            return voices[0]
+        raise ValueError("No voices available")
+    except Exception as e:
+        sinapsis_logger.error(f"Error getting default voice: {e}")
+        raise
+
+
+def load_input_text(input_data: list[TextPacket]) -> str:
+    """Loads and concatenates the text content from a list of TextPacket objects."""
+    return "".join([item.content for item in input_data])

{sinapsis_speech-0.3.5 → sinapsis_speech-0.4.0}/packages/sinapsis_elevenlabs/src/sinapsis_elevenlabs/templates/__init__.py

@@ -7,6 +7,8 @@ _root_lib_path = "sinapsis_elevenlabs.templates"
 _template_lookup = {
     "ElevenLabsTTS": f"{_root_lib_path}.elevenlabs_tts",
     "ElevenLabsVoiceGeneration": f"{_root_lib_path}.elevenlabs_voice_generation",
+    "ElevenLabsVoiceClone": f"{_root_lib_path}.elevenlabs_voice_clone",
+    "ElevenLabsSTS": f"{_root_lib_path}.elevenlabs_sts",
 }
 
 

{sinapsis_speech-0.3.5 → sinapsis_speech-0.4.0}/packages/sinapsis_elevenlabs/src/sinapsis_elevenlabs/templates/elevenlabs_base.py

@@ -5,10 +5,10 @@ import abc
 import os
 import uuid
 from io import BytesIO
-from typing import IO, Iterator, Literal
+from typing import IO, Iterable, Iterator, Literal
 
 from elevenlabs import Voice, VoiceSettings, save
-from elevenlabs.client import ElevenLabs, VoiceId, VoiceName
+from elevenlabs.client import ElevenLabs
 from elevenlabs.types import OutputFormat
 from pydantic import Field
 from sinapsis_core.data_containers.data_packet import AudioPacket, DataContainer, Packet
@@ -53,7 +53,7 @@ class ElevenLabsBase(Template, abc.ABC):
                 "mp3_44100_192", "pcm_16000", "pcm_22050", "pcm_24000", "pcm_44100", "ulaw_8000"]
             output_folder (str): The folder where generated audio files will be saved.
             stream (bool): If True, the audio is returned as a stream; otherwise, saved to a file.
-            voice (VoiceId | VoiceName | Voice): The voice to use for speech synthesis. This can be a voice ID (str),
+            voice (str | Voice | None): The voice to use for speech synthesis. This can be a voice ID (str),
                 a voice name (str) or an elevenlabs voice object (Voice).
             voice_settings (VoiceSettings): A dictionary of settings that control the behavior of the voice.
                 - stability (float)
@@ -76,7 +76,7 @@ class ElevenLabsBase(Template, abc.ABC):
         output_format: OutputFormat = "mp3_44100_128"
         output_folder: str = os.path.join(SINAPSIS_CACHE_DIR, "elevenlabs", "audios")
         stream: bool = False
-        voice: VoiceId | VoiceName | Voice = None
+        voice: str | Voice | None = None
         voice_settings: VoiceSettings = Field(default_factory=dict)  # type: ignore[arg-type]
 
     UIProperties = UIPropertiesMetadata(category="Elevenlabs", output_type=OutputTypes.AUDIO)
@@ -100,7 +100,7 @@ class ElevenLabsBase(Template, abc.ABC):
     def synthesize_speech(self, input_data: list[Packet]) -> RESPONSE_TYPE:
         """Abstract method for ElevenLabs speech synthesis."""
 
-    def _save_audio(self, response: Iterator[bytes] | bytes, file_format: str, idx: int) -> str:
+    def _save_audio(self, response: Iterable | bytes, file_format: str, idx: int) -> str:
         """Saves the audio to a file and returns the file path."""
         if self.attributes.output_file_name:
             file_name = self.attributes.output_file_name + "_" + str(idx)
@@ -116,7 +116,7 @@ class ElevenLabsBase(Template, abc.ABC):
             self.logger.error(f"File system error while saving speech to file: {e}")
             raise
 
-    def _generate_audio_stream(self, response: Iterator[bytes] | bytes) -> IO[bytes]:
+    def _generate_audio_stream(self, response: Iterable | bytes) -> IO[bytes]:
         """Generates and returns the audio stream."""
         audio_stream = BytesIO()
         try:
@@ -139,7 +139,7 @@ class ElevenLabsBase(Template, abc.ABC):
             self.logger.error(f"Value error while processing audio chunks: {e}")
             raise
 
-    def _process_audio_output(self, idx: int, response: Iterator[bytes] | bytes) -> str | IO[bytes]:
+    def _process_audio_output(self, idx: int, response: Iterable | bytes) -> str | IO[bytes]:
         """Processes a single audio output (either stream or file)."""
         if self.attributes.stream:
             return self._generate_audio_stream(response)

sinapsis_speech-0.4.0/packages/sinapsis_elevenlabs/src/sinapsis_elevenlabs/templates/elevenlabs_sts.py

@@ -0,0 +1,95 @@
+# -*- coding: utf-8 -*-
+"""Speech-To-Speech template for ElevenLabs."""
+
+from typing import Callable, Iterator, Literal
+
+from sinapsis_core.data_containers.data_packet import AudioPacket
+
+from sinapsis_elevenlabs.helpers.voice_utils import create_voice_settings, get_voice_id
+from sinapsis_elevenlabs.templates.elevenlabs_base import ElevenLabsBase
+
+
+class ElevenLabsSTS(ElevenLabsBase):
+    """Template to interact with the ElevenLabs Speech-to-Speech API.
+
+    This template takes an input audio and converts it to a new voice using
+    the ElevenLabs Speech-to-Speech (STS) API.
+
+    Usage example:
+
+    agent:
+      name: my_test_agent
+    templates:
+    - template_name: InputTemplate
+      class_name: InputTemplate
+      attributes: {}
+    - template_name: ElevenLabsSTS
+      class_name: ElevenLabsSTS
+      template_input: InputTemplate
+      attributes:
+        api_key: null
+        model: eleven_multilingual_sts_v2
+        output_file_name: null
+        output_format: mp3_44100_128
+        output_folder: ~/.cache/sinapsis/elevenlabs/audios
+        stream: false
+        voice: null
+        voice_settings:
+            stability: null
+            similarity_boost: null
+            style: null
+            use_speaker_boost: null
+            speed: null
+        streaming_latency: null
+
+    """
+
+    PACKET_TYPE_NAME: str = "audios"
+
+    class AttributesBaseModel(ElevenLabsBase.AttributesBaseModel):
+        """Attributes specific to ElevenLabs STS API interaction.
+
+        Attributes:
+            model (Literal): The STS model to use. Options are "eleven_english_sts_v2" or "eleven_multilingual_sts_v2".
+            streaming_latency (int | None): Optional latency optimization for streaming. Defaults to None.
+        """
+
+        model: Literal["eleven_english_sts_v2", "eleven_multilingual_sts_v2"] = "eleven_multilingual_sts_v2"
+        streaming_latency: int | None = None
+
+    def synthesize_speech(self, input_data: list[AudioPacket]) -> Iterator[bytes]:
+        """Sends an audio input to the ElevenLabs API for speech-to-speech synthesis.
+
+        Args:
+            input_data (list[AudioPacket]): List of AudioPacket objects containing the audio to be converted.
+                Only the first AudioPacket in the list is used.
+
+        Returns:
+            Iterator[bytes]: An iterator yielding audio data chunks in the output format specified.
+
+        Raises:
+            ValueError: If there is a problem with the input data or parameters.
+            TypeError: If the input data or files are of incorrect type.
+            KeyError: If the expected key is missing in the API response.
+        """
+        try:
+            method: Callable[..., Iterator[bytes]] = (
+                self.client.speech_to_speech.stream if self.attributes.stream else self.client.speech_to_speech.convert
+            )
+            return method(
+                voice_id=get_voice_id(self.client, voice=self.attributes.voice),
+                audio=input_data[0].content,
+                model_id=self.attributes.model,
+                voice_settings=create_voice_settings(self.attributes.voice_settings, as_json=True),
+                output_format=self.attributes.output_format,
+                optimize_streaming_latency=self.attributes.streaming_latency,
+            )
+        except ValueError as e:
+            self.logger.error(f"Value error synthesizing speech: {e}")
+            raise
+        except TypeError as e:
+            self.logger.error(f"Type error in input data or parameters: {e}")
+            raise
+        except KeyError as e:
+            self.logger.error(f"Missing key in input data or settings: {e}")
+            raise

{sinapsis_speech-0.3.5 → sinapsis_speech-0.4.0}/packages/sinapsis_elevenlabs/src/sinapsis_elevenlabs/templates/elevenlabs_tts.py

@@ -1,12 +1,13 @@
 # -*- coding: utf-8 -*-
 """Text-To-Speech template for ElevenLabs"""
 
-from typing import Iterator, Literal
+from typing import Callable, Iterator, Literal
 
 from sinapsis_core.data_containers.data_packet import TextPacket
 
 from sinapsis_elevenlabs.helpers.voice_utils import (
     create_voice_settings,
+    get_voice_id,
     load_input_text,
 )
 from sinapsis_elevenlabs.templates.elevenlabs_base import ElevenLabsBase
@@ -64,16 +65,16 @@ class ElevenLabsTTS(ElevenLabsBase):
         """
         input_text: str = load_input_text(input_data)
         try:
-            response: Iterator[bytes] = self.client.generate(
+            method: Callable[..., Iterator[bytes]] = (
+                self.client.text_to_speech.stream if self.attributes.stream else self.client.text_to_speech.convert
+            )
+            return method(
                 text=input_text,
-                voice=self.attributes.voice,
-                model=self.attributes.model,
+                voice_id=get_voice_id(self.client, self.attributes.voice),
+                model_id=self.attributes.model,
                 voice_settings=create_voice_settings(self.attributes.voice_settings),
                 output_format=self.attributes.output_format,
-                stream=self.attributes.stream,
             )
-
-            return response
         except ValueError as e:
             self.logger.error(f"Value error synthesizing speech: {e}")
             raise

sinapsis_speech-0.4.0/packages/sinapsis_elevenlabs/src/sinapsis_elevenlabs/templates/elevenlabs_voice_clone.py

@@ -0,0 +1,123 @@
+# -*- coding: utf-8 -*-
+"""Text-To-Speech template for ElevenLabs Voice Cloning."""
+
+from elevenlabs import Voice
+from sinapsis_core.data_containers.data_packet import AudioPacket, DataContainer
+
+from sinapsis_elevenlabs.templates.elevenlabs_tts import ElevenLabsTTS
+
+
+class ElevenLabsVoiceClone(ElevenLabsTTS):
+    """Template to clone a voice using the ElevenLabs API.
+
+    This template allows you to create a new custom voice in ElevenLabs by providing
+    one or more audio samples. The cloned voice can then be used for subsequent
+    text-to-speech synthesis within the Sinapsis pipeline.
+
+    Usage example:
+
+    agent:
+    name: my_test_agent
+    templates:
+    - template_name: InputTemplate
+      class_name: InputTemplate
+      attributes: {}
+    - template_name: ElevenLabsVoiceClone
+      class_name: ElevenLabsVoiceClone
+      template_input: InputTemplate
+      attributes:
+        api_key: null
+        model: eleven_turbo_v2_5
+        output_file_name: null
+        output_format: mp3_44100_128
+        output_folder: ~/.cache/sinapsis/elevenlabs/audios
+        stream: false
+        voice: null
+        voice_settings:
+            stability: null
+            similarity_boost: null
+            style: null
+            use_speaker_boost: null
+            speed: null
+        name: null
+        description: null
+        remove_background_noise: false
+
+    """
+
+    class AttributesBaseModel(ElevenLabsTTS.AttributesBaseModel):
+        """Attributes specific to the ElevenLabsVoiceClone class.
+
+        Attributes:
+            name (str | None): Name for the cloned voice. If None, a default name may be used.
+            description (str | None): Description for the cloned voice. Optional.
+            remove_background_noise (bool): Whether to remove background noise from samples. Defaults to False.
+        """
+
+        name: str | None = None
+        description: str | None = None
+        remove_background_noise: bool = False
+
+    def clone_voice(self, input_data: list[AudioPacket]) -> Voice:
+        """Clones a voice using the provided audio files.
+
+        Args:
+            input_data (list[AudioPacket]): List of AudioPacket objects containing the audio samples
+                to be used for voice cloning. Each AudioPacket's `content` should be a file-like object
+                or bytes representing the audio data.
+                **NOTE:** All provided audio packets are used as reference for a single cloned voice.
+
+        Returns:
+            Voice: The cloned Voice object as returned by the ElevenLabs API.
+
+        Raises:
+            ValueError: If there is a problem with the input data or parameters.
+            TypeError: If the input data or files are of incorrect type.
+            KeyError: If the expected key is missing in the API response.
+        """
+        files = [audio.content for audio in input_data]
+        try:
+            clone_response = self.client.voices.ivc.create(
+                name=self.attributes.name,
+                files=files,
+                description=self.attributes.description,
+                remove_background_noise=self.attributes.remove_background_noise,
+            )
+            cloned_voice = self.client.voices.get(clone_response.voice_id)
+            self.logger.info(f"Voice cloned successfully with IVC: {cloned_voice.name}")
+            return cloned_voice
+        except ValueError as e:
+            self.logger.error(f"Value error in input data or parameters: {e}")
+            raise
+        except TypeError as e:
+            self.logger.error(f"Type error with input data or files: {e}")
+            raise
+        except KeyError as e:
+            self.logger.error(f"Missing expected key in API response: {e}")
+            raise
+
+    def execute(self, container: DataContainer) -> DataContainer:
+        """Executes the voice cloning process and generates the speech output.
+
+        Args:
+            container (DataContainer): The input DataContainer, expected to contain
+                one or more AudioPacket objects in the `audios` attribute.
+
+        Returns:
+            DataContainer: The updated DataContainer. If cloning is successful,
+                the cloned voice is set in `self.attributes.voice` and the parent
+                TTS execution is performed using the new voice.
+
+        Side Effects:
+            - Updates `self.attributes.voice` with the cloned Voice object.
+            - May log errors or info messages.
+        """
+        audios = container.audios
+        if not audios:
+            self.logger.debug("No audios provided to clone voice")
+            return container
+        self.attributes.voice = self.clone_voice(audios)
+
+        container = super().execute(container)
+
+        return container

sinapsis_speech-0.4.0/packages/sinapsis_orpheus_cpp/src/sinapsis_orpheus_cpp/templates/__init__.py

@@ -0,0 +1,20 @@
+# -*- coding: utf-8 -*-
+import importlib
+from typing import Callable
+
+_root_lib_path = "sinapsis_orpheus_cpp.templates"
+
+_template_lookup = {
+    "OrpheusTTS": f"{_root_lib_path}.orpheus_tts",
+}
+
+
+def __getattr__(name: str) -> Callable:
+    if name in _template_lookup:
+        module = importlib.import_module(_template_lookup[name])
+        return getattr(module, name)
+
+    raise AttributeError(f"template `{name}` not found in {_root_lib_path}")
+
+
+__all__ = list(_template_lookup.keys())