phonexia-enhanced-speech-to-text-built-on-whisper-client 1.3.0__tar.gz → 1.5.0__tar.gz

{phonexia_enhanced_speech_to_text_built_on_whisper_client-1.3.0 → phonexia_enhanced_speech_to_text_built_on_whisper_client-1.5.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: phonexia-enhanced-speech-to-text-built-on-whisper-client
-Version: 1.3.0
+Version: 1.5.0
 Summary: Client for communication with Phonexia Enhanced Speech To Text Built On Whisper microservice.
 Keywords: grpc,transcription,STT,ASR,speech to text,speech,language,microservice
 Author: Phonexia
@@ -13,7 +13,10 @@ Classifier: Programming Language :: Python :: 3.10
 Classifier: Programming Language :: Python :: 3.11
 Classifier: Programming Language :: Python :: 3.12
 Requires-Dist: grpcio (>=1.54.0,<2.0.0)
+Requires-Dist: numpy (<2.0.0) ; python_version < "3.12"
+Requires-Dist: numpy (>=2.0.0) ; python_version >= "3.12"
 Requires-Dist: phonexia-grpc (>=2.0.0,<3.0.0)
+Requires-Dist: soundfile (>=0.12.1,<0.13.0)
 Project-URL: Homepage, https://phonexia.com
 Project-URL: Issues, https://phonexia.atlassian.net/servicedesk/customer/portal/15/group/20/create/40
 Project-URL: protofiles, https://github.com/phonexia/protofiles

{phonexia_enhanced_speech_to_text_built_on_whisper_client-1.3.0 → phonexia_enhanced_speech_to_text_built_on_whisper_client-1.5.0}/phonexia_enhanced_speech_to_text_built_on_whisper_client.py

@@ -2,15 +2,16 @@ import argparse
 import json
 import logging
 import os
+from datetime import datetime
 from enum import Enum
 from typing import Iterator, Optional
 
 import google.protobuf.duration_pb2
 import grpc
-from google.protobuf.json_format import MessageToDict
-
 import phonexia.grpc.technologies.enhanced_speech_to_text_built_on_whisper.v1.enhanced_speech_to_text_built_on_whisper_pb2_grpc as stt_grpc
-from phonexia.grpc.common.core_pb2 import Audio, TimeRange
+import soundfile
+from google.protobuf.json_format import MessageToDict
+from phonexia.grpc.common.core_pb2 import Audio, RawAudioConfig, TimeRange
 from phonexia.grpc.technologies.enhanced_speech_to_text_built_on_whisper.v1.enhanced_speech_to_text_built_on_whisper_pb2 import (
     TranscribeConfig,
     TranscribeRequest,
@@ -44,19 +45,42 @@ def transcribe_request_iterator(
     start: Optional[float],
     end: Optional[float],
     enable_language_switching: bool = False,
+    use_raw_audio: bool = False,
 ) -> Iterator[TranscribeRequest]:
     time_range = TimeRange(start=time_to_duration(start), end=time_to_duration(end))
     config = TranscribeConfig(
         language=specified_language, enable_language_switching=enable_language_switching
     )
 
-    with open(file, "rb") as f:
-        while chunk := f.read(CHUNK_SIZE):
-            yield TranscribeRequest(
-                audio=Audio(content=chunk, time_range=time_range), config=config
+    if use_raw_audio:
+        with soundfile.SoundFile(file) as r:
+            raw_audio_config = RawAudioConfig(
+                channels=r.channels,
+                sample_rate_hertz=r.samplerate,
+                encoding=RawAudioConfig.AudioEncoding.PCM16,
             )
-            time_range = None
-            config = None
+
+            for data in r.blocks(blocksize=r.samplerate, dtype="int16"):
+                logging.debug("Sending chunk of size %d samples", len(data))
+                yield TranscribeRequest(
+                    audio=Audio(
+                        content=data.flatten().tobytes(),
+                        time_range=time_range,
+                        raw_audio_config=raw_audio_config,
+                    ),
+                    config=config,
+                )
+                time_range = None
+                raw_audio_config = None
+                config = None
+    else:
+        with open(file, "rb") as f:
+            while chunk := f.read(CHUNK_SIZE):
+                yield TranscribeRequest(
+                    audio=Audio(content=chunk, time_range=time_range), config=config
+                )
+                time_range = None
+                config = None
 
 
 def translate_request_iterator(
@@ -87,6 +111,7 @@ def transcribe(
     metadata: Optional[list],
     task: Task,
     enable_language_switching: bool = False,
+    use_raw_audio: bool = False,
 ):
     stub = stt_grpc.SpeechToTextStub(channel)
     if task == Task.transcribe:
@@ -97,6 +122,7 @@ def transcribe(
                 start=start,
                 end=end,
                 enable_language_switching=enable_language_switching,
+                use_raw_audio=use_raw_audio,
             ),
             metadata=metadata,
         )
@@ -197,6 +223,7 @@ def main():
         help="Enable dynamic language switching during transcription, with the language being detected approximately every 30 seconds",
     )
     parser.add_argument("file", type=str, help="Path to input file")
+    parser.add_argument("--use_raw_audio", action="store_true", help="Send a raw audio in")
 
     args = parser.parse_args()
 
@@ -227,6 +254,8 @@ def main():
             else grpc.insecure_channel(target=args.host)
         )
 
+        start_time = datetime.now()
+
         transcribe(
             channel=channel,
             file=args.file,
@@ -236,8 +265,11 @@ def main():
             metadata=args.metadata,
             task=args.task,
             enable_language_switching=args.enable_language_switching,
+            use_raw_audio=args.use_raw_audio,
         )
 
+        logging.debug(f"Elapsed time {(datetime.now() - start_time)}")
+
     except grpc.RpcError:
         logging.exception("RPC failed")
         exit(1)

{phonexia_enhanced_speech_to_text_built_on_whisper_client-1.3.0 → phonexia_enhanced_speech_to_text_built_on_whisper_client-1.5.0}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "phonexia-enhanced-speech-to-text-built-on-whisper-client"
-version = "1.3.0"
+version = "1.5.0"
 description = "Client for communication with Phonexia Enhanced Speech To Text Built On Whisper microservice."
 readme = "pypi-README.md"
 keywords = ["grpc", "transcription", "STT", "ASR", "speech to text", "speech", "language", "microservice"]
@@ -18,6 +18,11 @@ enhanced_speech_to_text_built_on_whisper_client = 'phonexia_enhanced_speech_to_t
 python = ">=3.8,<4.0"
 grpcio = "^1.54.0"
 phonexia-grpc = {version="^2.0.0", source="pypi"}
+soundfile = "^0.12.1"
+numpy = [
+    { version = "<2.0.0", markers = "python_version < '3.12'" },
+    { version = ">=2.0.0", markers = "python_version >= '3.12'" }
+]
 
 [tool.poetry.group.dev.dependencies]
 pytest = "^8.0.0"
@@ -28,14 +33,12 @@ black = "^24.0.0"
 ruff = "^0.4.0"
 
 [[tool.poetry.source]]
-name = "gitlab"
-url = "https://gitlab.cloud.phonexia.com/api/v4/groups/39/-/packages/pypi/simple"
+name = "PyPI"
 priority = "primary"
 
-
 [[tool.poetry.source]]
-name = "PyPI"
-priority = "default"
+name = "gitlab"
+url = "https://gitlab.cloud.phonexia.com/api/v4/groups/39/-/packages/pypi/simple"
 
 [build-system]
 requires = ["poetry-core>=1.0.0"]