sinapsis-speech 0.4.3__py3-none-any.whl → 0.4.4__py3-none-any.whl

sinapsis_elevenlabs/src/sinapsis_elevenlabs/templates/elevenlabs_base.py

@@ -2,7 +2,6 @@
 """Base template for ElevenLabs speech synthesis"""
 
 import abc
-import os
 from typing import Generator, Iterable, Iterator, Literal
 
 import numpy as np
@@ -18,7 +17,6 @@ from sinapsis_core.template_base.base_models import (
     UIPropertiesMetadata,
 )
 from sinapsis_core.template_base.template import Template
-from sinapsis_core.utils.env_var_keys import WORKING_DIR
 from sinapsis_generic_data_tools.helpers.audio_encoder import audio_bytes_to_numpy
 
 from sinapsis_elevenlabs.helpers.env_var_keys import ELEVENLABS_API_KEY
@@ -47,8 +45,7 @@ class ElevenLabsBase(Template, abc.ABC):
         Args:
             api_key (str): The API used key to authenticate with ElevenLabs' API.
             model (Literal): The model identifier to use for speech synthesis.
-            output_file_name (str | None): Optional name for saved audio file.
-                If not provided a random UUI will be used as file name. Defaults to None.
+
             output_format (OutputFormat): The output audio format and quality. Options include:
                 ["mp3_22050_32", "mp3_44100_32", "mp3_44100_64", "mp3_44100_96", "mp3_44100_128",
                 "mp3_44100_192", "pcm_16000", "pcm_22050", "pcm_24000", "pcm_44100", "ulaw_8000"]
@@ -71,9 +68,7 @@ class ElevenLabsBase(Template, abc.ABC):
             "eleven_english_sts_v2",
             "eleven_multilingual_sts_v2",
         ] = "eleven_turbo_v2_5"
-        output_file_name: str | None = None
         output_format: OutputFormat = "mp3_44100_128"
-        output_folder: str = os.path.join(WORKING_DIR, "elevenlabs", "audios")
         stream: bool = False
         voice: str | Voice | None = None
         voice_settings: VoiceSettings = Field(default_factory=dict)  # type: ignore[arg-type]

sinapsis_elevenlabs/src/sinapsis_elevenlabs/templates/elevenlabs_sts.py

@@ -33,9 +33,7 @@ class ElevenLabsSTS(ElevenLabsBase):
       attributes:
         api_key: null
         model: eleven_multilingual_sts_v2
-        output_file_name: null
         output_format: mp3_44100_128
-        output_folder: <WORKING_DIR>/elevenlabs/audios
         stream: false
         voice: null
         voice_settings:

sinapsis_elevenlabs/src/sinapsis_elevenlabs/templates/elevenlabs_tts.py

@@ -40,7 +40,6 @@ class ElevenLabsTTS(ElevenLabsBase):
         voice_settings: null
         model: eleven_turbo_v2_5
         output_format: mp3_44100_128
-        output_folder: <WORKING_DIR>/elevenlabs/audios
         stream: false
 
     """

sinapsis_elevenlabs/src/sinapsis_elevenlabs/templates/elevenlabs_voice_clone.py

@@ -32,9 +32,7 @@ class ElevenLabsVoiceClone(ElevenLabsTTS):
       attributes:
         api_key: null
         model: eleven_turbo_v2_5
-        output_file_name: null
         output_format: mp3_44100_128
-        output_folder: <WORKING_DIR>/elevenlabs/audios
         stream: false
         voice: null
         voice_settings:

sinapsis_elevenlabs/src/sinapsis_elevenlabs/templates/elevenlabs_voice_generation.py

@@ -37,7 +37,6 @@ class ElevenLabsVoiceGeneration(ElevenLabsBase):
         voice_settings: null
         model: eleven_turbo_v2_5
         output_format: mp3_44100_128
-        output_folder: <WORKING_DIR>/elevenlabs/audios
         stream: false
         voice_description: An old British male with a raspy, deep voice. Professional,
           relaxed and assertive

sinapsis_f5_tts/src/sinapsis_f5_tts/templates/f5_tts_inference.py

@@ -14,7 +14,13 @@ from sinapsis_core.data_containers.data_packet import (
     DataContainer,
 )
 from sinapsis_core.template_base import Template
-from sinapsis_core.template_base.base_models import OutputTypes, TemplateAttributes, UIPropertiesMetadata
+from sinapsis_core.template_base.base_models import (
+    OutputTypes,
+    TemplateAttributes,
+    TemplateAttributeType,
+    UIPropertiesMetadata,
+)
+from sinapsis_core.utils.env_var_keys import SINAPSIS_CACHE_DIR
 
 from sinapsis_f5_tts.helpers.tags import Tags
 
@@ -107,6 +113,8 @@ class F5TTSInferenceAttributes(TemplateAttributes):
 
     device: str | None = Field(default=None, json_schema_extra={F5CliKeys.cli_param: "--device"})
 
+    root_dir: str | None = None
+
 
 class F5TTSInference(Template):
     """Template for performing text-to-speech synthesis using the F5TTS model.
@@ -155,6 +163,11 @@ class F5TTSInference(Template):
         tags=[Tags.AUDIO, Tags.AUDIO_GENERATION, Tags.F5TTS, Tags.SPEECH, Tags.TEXT_TO_SPEECH],
     )
 
+    def __init__(self, attributes: TemplateAttributeType) -> None:
+        super().__init__(attributes)
+        self.attributes.root_dir = self.attributes.root_dir or SINAPSIS_CACHE_DIR
+        self.attributes.ref_audio = os.path.join(self.attributes.root_dir, self.attributes.ref_audio)
+
     def _add_attribute_to_command(self, cli_command: list[str], field_name: str, field: Any) -> None:
         """
         This method examines each attribute field's metadata to determine if and how

sinapsis_parakeet_tdt/src/sinapsis_parakeet_tdt/templates/parakeet_tdt.py

@@ -15,6 +15,7 @@ from sinapsis_core.template_base.base_models import (
     UIPropertiesMetadata,
 )
 from sinapsis_core.template_base.template import Template
+from sinapsis_core.utils.env_var_keys import SINAPSIS_CACHE_DIR
 
 from sinapsis_parakeet_tdt.helpers.tags import Tags
 
@@ -36,6 +37,7 @@ class ParakeetTDTInferenceAttributes(TemplateAttributes):
 
     model_name: str = "nvidia/parakeet-tdt-0.6b-v2"
     audio_paths: list[str] | None = None
+    root_dir: str | None = None
     enable_timestamps: bool = False
     timestamp_level: Literal["char", "word", "segment"] = "word"
     device: Literal["cpu", "cuda"] = "cuda"
@@ -88,6 +90,7 @@ class ParakeetTDTInference(Template):
 
     def __init__(self, attributes: TemplateAttributes) -> None:
         super().__init__(attributes)
+        self.attributes.root_dir = self.attributes.root_dir or SINAPSIS_CACHE_DIR
         self._load_model()
 
     def _load_model(self) -> None:
@@ -131,9 +134,10 @@ class ParakeetTDTInference(Template):
         """
         sources = []
         for path in paths:
-            if not os.path.exists(path):
-                self.logger.warning(f"Audio file not found: {path}")
-            sources.append(path)
+            full_path = os.path.join(self.attributes.root_dir, path)
+            if not os.path.exists(full_path):
+                self.logger.warning(f"Audio file not found: {full_path}")
+            sources.append(full_path)
         return sources
 
     def get_audio_sources(self, container: DataContainer) -> list[str]:
@@ -156,7 +160,6 @@ class ParakeetTDTInference(Template):
 
         if not sources and self.attributes.audio_paths:
             sources = self.get_sources_from_paths(self.attributes.audio_paths)
-
         return sources
 
     @staticmethod
@@ -220,6 +223,7 @@ class ParakeetTDTInference(Template):
         Returns:
             list[Any]: List of transcription results from the ASR model.
         """
+
         return self.model.transcribe(
             sources,
             timestamps=self.attributes.enable_timestamps,

{sinapsis_speech-0.4.3.dist-info → sinapsis_speech-0.4.4.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: sinapsis-speech
-Version: 0.4.3
+Version: 0.4.4
 Summary: Generate speech using various libraries.
 Author-email: SinapsisAI <dev@sinapsis.tech>
 Project-URL: Homepage, https://sinapsis.tech

{sinapsis_speech-0.4.3.dist-info → sinapsis_speech-0.4.4.dist-info}/RECORD

@@ -4,16 +4,16 @@ sinapsis_elevenlabs/src/sinapsis_elevenlabs/helpers/env_var_keys.py,sha256=j8J64
 sinapsis_elevenlabs/src/sinapsis_elevenlabs/helpers/tags.py,sha256=EzEwYJqwPFhSzJB7K8g1HGm3xiy6M_kE1j19TYQAfS8,402
 sinapsis_elevenlabs/src/sinapsis_elevenlabs/helpers/voice_utils.py,sha256=2Ym4suCk8wy-Nj2Hmk0uu3_-3nu1QlSs_KubDydm5wY,3383
 sinapsis_elevenlabs/src/sinapsis_elevenlabs/templates/__init__.py,sha256=UG35_hown3HITVR42iK_e3yVsUbuq2oYTLpCGwJ89L4,708
-sinapsis_elevenlabs/src/sinapsis_elevenlabs/templates/elevenlabs_base.py,sha256=Gb0wTth3GAAN_22RLm0mPPXtw3eUd2DJQVRc4itEqAM,7900
-sinapsis_elevenlabs/src/sinapsis_elevenlabs/templates/elevenlabs_sts.py,sha256=72WKvjxcro8PD234iFBV9kiYIBLTQUPU25xzGwbPjv8,3799
-sinapsis_elevenlabs/src/sinapsis_elevenlabs/templates/elevenlabs_tts.py,sha256=CePqHzLHrsgKF87nhBQfZE2htB6DEE2iP7_LHjFUQ_E,3047
-sinapsis_elevenlabs/src/sinapsis_elevenlabs/templates/elevenlabs_voice_clone.py,sha256=pLq3U1ZZSW1NUOJEhvUJHhSxjEQiEKFiqHJikhOjye8,4967
-sinapsis_elevenlabs/src/sinapsis_elevenlabs/templates/elevenlabs_voice_generation.py,sha256=CvPOrdLBA4XNA-G1XYscT2eUvPa-elJNuNHvm8bzhOk,3078
+sinapsis_elevenlabs/src/sinapsis_elevenlabs/templates/elevenlabs_base.py,sha256=dry8bhtdn1KBZroPoUAqDHTuXlSqrmD549DEe-HcEXw,7542
+sinapsis_elevenlabs/src/sinapsis_elevenlabs/templates/elevenlabs_sts.py,sha256=PaWfAPyo0eTjHxE3plEzQ4Rk5m88Q_2Nqp21yyJLHTQ,3713
+sinapsis_elevenlabs/src/sinapsis_elevenlabs/templates/elevenlabs_tts.py,sha256=8yJwh-COxotjMTrQln85HpE7-znq6GSs4_qHlod9U_g,2992
+sinapsis_elevenlabs/src/sinapsis_elevenlabs/templates/elevenlabs_voice_clone.py,sha256=YEtT-z6qv8hwAAwrNIZHBpV3vhvgUVaSbgpf3bm8_ac,4881
+sinapsis_elevenlabs/src/sinapsis_elevenlabs/templates/elevenlabs_voice_generation.py,sha256=VTgquS9l9jimVtaDG62TCKYC8xXCFqwcJVueeCQXSpQ,3023
 sinapsis_f5_tts/src/sinapsis_f5_tts/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 sinapsis_f5_tts/src/sinapsis_f5_tts/helpers/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 sinapsis_f5_tts/src/sinapsis_f5_tts/helpers/tags.py,sha256=T9RbgsPgq7Wc-0Lu3W8Si4bxjQsyWbX6hPrc-dakdCs,208
 sinapsis_f5_tts/src/sinapsis_f5_tts/templates/__init__.py,sha256=28BOPAr9GG1jYcrXi45ZWO1n2FAZJOdDcmRkOXdEYmk,496
-sinapsis_f5_tts/src/sinapsis_f5_tts/templates/f5_tts_inference.py,sha256=HooFxYB1lqRKuzGjMDiw48Xzm34YI81dE0uD3WWm89A,16344
+sinapsis_f5_tts/src/sinapsis_f5_tts/templates/f5_tts_inference.py,sha256=OBAWVOg_QId14hbftEa_oJHap6jpqAZeVj8ZnYN0Vsk,16774
 sinapsis_kokoro/src/sinapsis_kokoro/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 sinapsis_kokoro/src/sinapsis_kokoro/helpers/kokoro_utils.py,sha256=2IMJuwURPKK7keIkgS-rpGD28REG5M1FwW0COGcm3nI,1573
 sinapsis_kokoro/src/sinapsis_kokoro/helpers/tags.py,sha256=AiHIvqmujKX6tYQ4lEXjRGhq8Ujst8gZwVmcAjS7u3k,210
@@ -29,8 +29,8 @@ sinapsis_parakeet_tdt/src/sinapsis_parakeet_tdt/__init__.py,sha256=47DEQpj8HBSa-
 sinapsis_parakeet_tdt/src/sinapsis_parakeet_tdt/helpers/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 sinapsis_parakeet_tdt/src/sinapsis_parakeet_tdt/helpers/tags.py,sha256=OKZbq4zIL6XWM7eG5WuQ3dWYkmYNWjuCnlseXmjR_j0,262
 sinapsis_parakeet_tdt/src/sinapsis_parakeet_tdt/templates/__init__.py,sha256=3LppgbS6v70Rmx__yXXQgnoZ2ZBHcXkXeWZYQQf6Zwg,504
-sinapsis_parakeet_tdt/src/sinapsis_parakeet_tdt/templates/parakeet_tdt.py,sha256=ECNDd_zxkkop3WikQpkRW_bjLAMjROGpdWbChvaJdBE,9948
-sinapsis_speech-0.4.3.dist-info/licenses/LICENSE,sha256=hIahDEOTzuHCU5J2nd07LWwkLW7Hko4UFO__ffsvB-8,34523
+sinapsis_parakeet_tdt/src/sinapsis_parakeet_tdt/templates/parakeet_tdt.py,sha256=Tw9S8Nqf74lXwUxBodaLK_JaQvh9ITt8cWFQJ2QNP6s,10210
+sinapsis_speech-0.4.4.dist-info/licenses/LICENSE,sha256=hIahDEOTzuHCU5J2nd07LWwkLW7Hko4UFO__ffsvB-8,34523
 sinapsis_zonos/src/sinapsis_zonos/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 sinapsis_zonos/src/sinapsis_zonos/helpers/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 sinapsis_zonos/src/sinapsis_zonos/helpers/tags.py,sha256=Y7MKQNx1E2k7ebF6r_1l1nBeS5k8hO424yFTT9NI7Rg,244
@@ -38,7 +38,7 @@ sinapsis_zonos/src/sinapsis_zonos/helpers/zonos_keys.py,sha256=m1GdOYfzP73JGmtxH
 sinapsis_zonos/src/sinapsis_zonos/helpers/zonos_tts_utils.py,sha256=bwu88wsJGzEqbssgb-wpS_7lFscJ74J8cgyca-hX_Qw,6422
 sinapsis_zonos/src/sinapsis_zonos/templates/__init__.py,sha256=A-_F0K3hbEFqeWWAh4YftgU9CFX-WHrauSiCAww9yp8,482
 sinapsis_zonos/src/sinapsis_zonos/templates/zonos_tts.py,sha256=h5EToXoJgAgjqvz9WLDfSjhCsV5zgBwZrX5cTJ4VnhM,7679
-sinapsis_speech-0.4.3.dist-info/METADATA,sha256=UZXiSYdXx8deyu9p28aaUkIDyxtfm1REJYrvV2JSj6E,12783
-sinapsis_speech-0.4.3.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-sinapsis_speech-0.4.3.dist-info/top_level.txt,sha256=KvdwXupt5wnqb_4XGRcuJaL9Glgdw-DBvRkNzhgl_Ds,110
-sinapsis_speech-0.4.3.dist-info/RECORD,,
+sinapsis_speech-0.4.4.dist-info/METADATA,sha256=dWpD72J-S4yNHvKEZY0dECGMfQ3wqX63x18iOf4bupw,12783
+sinapsis_speech-0.4.4.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+sinapsis_speech-0.4.4.dist-info/top_level.txt,sha256=KvdwXupt5wnqb_4XGRcuJaL9Glgdw-DBvRkNzhgl_Ds,110
+sinapsis_speech-0.4.4.dist-info/RECORD,,