dora-vad 0.3.10rc0__py3-none-any.whl → 0.3.11__py3-none-any.whl

dora_vad/__init__.py

@@ -1,3 +1,5 @@
+"""TODO: Add docstring."""
+
 import os
 
 # Define the path to the README file relative to the package directory

dora_vad/main.py

@@ -1,3 +1,5 @@
+"""TODO: Add docstring."""
+
 import os
 
 import numpy as np
@@ -7,14 +9,15 @@ from dora import Node
 from silero_vad import get_speech_timestamps, load_silero_vad
 
 model = load_silero_vad()
-MIN_SILENCE_DURATION_MS = int(os.getenv("MIN_SILENCE_DURATION_MS", "100"))
+MIN_SILENCE_DURATION_MS = int(os.getenv("MIN_SILENCE_DURATION_MS", "200"))
 MIN_SPEECH_DURATION_MS = int(os.getenv("MIN_SPEECH_DURATION_MS", "300"))
-
-MIN_AUDIO_SAMPLING_DURAION_S = int(os.getenv("MAX_AUDIO_DURATION_S", "20"))
-MAX_AUDIO_DURAION_S = int(os.getenv("MAX_AUDIO_DURATION_S", "75"))
+THRESHOLD = float(os.getenv("THRESHOLD", "0.4"))
+MAX_AUDIO_DURATION_S = float(os.getenv("MAX_AUDIO_DURATION_S", "75"))
+MIN_AUDIO_SAMPLING_DURATION_MS = int(os.getenv("MIN_AUDIO_SAMPLING_DURATION_MS", "500"))
 
 
 def main():
+    """TODO: Add docstring."""
     node = Node()
     last_audios = []
     while True:
@@ -26,10 +29,11 @@ def main():
             last_audios += [audio]
             last_audios = last_audios[-100:]
             audio = np.concatenate(last_audios)
+            sr = event["metadata"].get("sample_rate", 16000)
             speech_timestamps = get_speech_timestamps(
                 torch.from_numpy(audio),
                 model,
-                threshold=0.2,
+                threshold=THRESHOLD,
                 min_speech_duration_ms=MIN_SPEECH_DURATION_MS,
                 min_silence_duration_ms=MIN_SILENCE_DURATION_MS,
             )
@@ -37,16 +41,20 @@ def main():
             # Check ig there is timestamp
             if (
                 len(speech_timestamps) > 0
-                and len(last_audios) > MIN_AUDIO_SAMPLING_DURAION_S
+                and len(audio) > MIN_AUDIO_SAMPLING_DURATION_MS * sr / 1000
             ):
                 # Check if the audio is not cut at the end. And only return if there is a long time spent
                 if speech_timestamps[-1]["end"] == len(audio):
+                    node.send_output(
+                        "timestamp_start",
+                        pa.array([speech_timestamps[-1]["start"]]),
+                    )
                     continue
                 audio = audio[0 : speech_timestamps[-1]["end"]]
-                node.send_output("audio", pa.array(audio))
+                node.send_output("audio", pa.array(audio), metadata={"sample_rate": sr})
                 last_audios = [audio[speech_timestamps[-1]["end"] :]]
 
             # If there is no sound for too long return the audio
-            elif len(last_audios) > 75:
-                node.send_output("audio", pa.array(audio))
+            elif len(last_audios) > MAX_AUDIO_DURATION_S:
+                node.send_output("audio", pa.array(audio), metadata={"sample_rate": sr})
                 last_audios = []

{dora_vad-0.3.10rc0.dist-info → dora_vad-0.3.11.dist-info}/METADATA

@@ -1,12 +1,12 @@
-Metadata-Version: 2.2
+Metadata-Version: 2.4
 Name: dora-vad
-Version: 0.3.10rc0
+Version: 0.3.11
 Summary: Dora Node for Text translating using Argostranslate
 Author-email: Haixuan Xavier Tao <tao.xavier@outlook.com>
 License: MIT
 Requires-Python: >=3.8
 Description-Content-Type: text/markdown
-Requires-Dist: dora-rs>=0.3.6
+Requires-Dist: dora-rs>=0.3.9
 Requires-Dist: numpy<2.0.0
 Requires-Dist: silero-vad>=5.1
 

dora_vad-0.3.11.dist-info/RECORD

@@ -0,0 +1,7 @@
+dora_vad/__init__.py,sha256=tF7WHhHiDweUUzyHsbmFe_ktphE08aA5j33E4ja1udA,381
+dora_vad/main.py,sha256=LVtwdNkQH6j9SUn0LsQ3N3oWf4XnxR4zTEzIH7BdVFY,2360
+dora_vad-0.3.11.dist-info/METADATA,sha256=OOGdc9k4RtOlhJqRe4pDV2Dlvv4eByagSyrytJMftPU,1461
+dora_vad-0.3.11.dist-info/WHEEL,sha256=CmyFI0kx5cdEMTLiONQRbGQwjIoR1aIYB7eCAQ4KPJ0,91
+dora_vad-0.3.11.dist-info/entry_points.txt,sha256=X82-C0mpNH9bVjwn24mzfoWX5ssrK6W1Ylk1HPDR0ME,48
+dora_vad-0.3.11.dist-info/top_level.txt,sha256=1EUgMajnck6tHR4mFagv_uv-t2A6Cabd2wYtEldKm2Q,9
+dora_vad-0.3.11.dist-info/RECORD,,

{dora_vad-0.3.10rc0.dist-info → dora_vad-0.3.11.dist-info}/WHEEL

@@ -1,5 +1,5 @@
 Wheel-Version: 1.0
-Generator: setuptools (75.8.0)
+Generator: setuptools (78.1.0)
 Root-Is-Purelib: true
 Tag: py3-none-any
 

dora_vad-0.3.10rc0.dist-info/RECORD

@@ -1,7 +0,0 @@
-dora_vad/__init__.py,sha256=HuSK3dnyI9Pb5QAuaKFwQQ3J5SIZnLcKHPJO0norGzc,353
-dora_vad/main.py,sha256=HfdT87XdtlZfoNRfKbWKl-ylVNFoq9t8acfrzB6n3wE,1913
-dora_vad-0.3.10rc0.dist-info/METADATA,sha256=vd9g9MJob-CnGjULIw4r1xwRDZrMjYScFpt-cKzFIq8,1464
-dora_vad-0.3.10rc0.dist-info/WHEEL,sha256=In9FTNxeP60KnTkGw7wk6mJPYd_dQSjEZmXdBdMCI-8,91
-dora_vad-0.3.10rc0.dist-info/entry_points.txt,sha256=X82-C0mpNH9bVjwn24mzfoWX5ssrK6W1Ylk1HPDR0ME,48
-dora_vad-0.3.10rc0.dist-info/top_level.txt,sha256=1EUgMajnck6tHR4mFagv_uv-t2A6Cabd2wYtEldKm2Q,9
-dora_vad-0.3.10rc0.dist-info/RECORD,,