aadnk's picture
Fix merging of VAD segments
8781620
raw
history blame
13.7 kB
from abc import ABC, abstractmethod
from collections import Counter
from dis import dis
import re
from typing import Any, Iterator, List, Dict
from pprint import pprint
# Workaround for https://github.com/tensorflow/tensorflow/issues/48797
try:
import tensorflow as tf
except ModuleNotFoundError:
# Error handling
pass
import torch
import ffmpeg
import numpy as np
from src.utils import format_timestamp
# Defaults for Silero
# TODO: Make these configurable?
SPEECH_TRESHOLD = 0.3
MAX_SILENT_PERIOD = 10 # seconds
MAX_MERGE_SIZE = 150 # Do not create segments larger than 2.5 minutes
SEGMENT_PADDING_LEFT = 1 # Start detected text segment early
SEGMENT_PADDING_RIGHT = 1 # End detected segments late
# Whether to attempt to transcribe non-speech
TRANSCRIBE_NON_SPEECH = False
# Minimum size of segments to process
MIN_SEGMENT_DURATION = 1
class AbstractTranscription(ABC):
def __init__(self, segment_padding_left: int = None, segment_padding_right = None, max_silent_period: int = None, max_merge_size: int = None, transcribe_non_speech: bool = False):
self.sampling_rate = 16000
self.segment_padding_left = segment_padding_left
self.segment_padding_right = segment_padding_right
self.max_silent_period = max_silent_period
self.max_merge_size = max_merge_size
self.transcribe_non_speech = transcribe_non_speech
def get_audio_segment(self, str, start_time: str = None, duration: str = None):
return load_audio(str, self.sampling_rate, start_time, duration)
@abstractmethod
def get_transcribe_timestamps(self, audio: str):
"""
Get the start and end timestamps of the sections that should be transcribed by this VAD method.
Parameters
----------
audio: str
The audio file.
Returns
-------
A list of start and end timestamps, in fractional seconds.
"""
return
def transcribe(self, audio: str, whisperCallable):
"""
Transcribe the given audo file.
Parameters
----------
audio: str
The audio file.
whisperCallable: Callable[[Union[str, np.ndarray, torch.Tensor]], dict[str, Union[dict, Any]]]
The callback that is used to invoke Whisper on an audio file/buffer.
Returns
-------
A list of start and end timestamps, in fractional seconds.
"""
# get speech timestamps from full audio file
seconds_timestamps = self.get_transcribe_timestamps(audio)
padded = self.pad_timestamps(seconds_timestamps, self.segment_padding_left, self.segment_padding_right)
merged = self.merge_timestamps(padded, self.max_silent_period, self.max_merge_size)
print("Timestamps:")
pprint(merged)
if self.transcribe_non_speech:
max_audio_duration = float(ffmpeg.probe(audio)["format"]["duration"])
# Expand segments to include the gaps between them
merged = self.expand_gaps(merged, total_duration=max_audio_duration)
print("Transcribing non-speech:")
pprint(merged)
result = {
'text': "",
'segments': [],
'language': ""
}
languageCounter = Counter()
# For each time segment, run whisper
for segment in merged:
segment_start = segment['start']
segment_end = segment['end']
segment_expand_amount = segment.get('expand_amount', 0)
segment_duration = segment_end - segment_start
if segment_duration < MIN_SEGMENT_DURATION:
continue;
segment_audio = self.get_audio_segment(audio, start_time = str(segment_start), duration = str(segment_duration))
print("Running whisper from ", format_timestamp(segment_start), " to ", format_timestamp(segment_end), ", duration: ", segment_duration, "expanded: ", segment_expand_amount)
segment_result = whisperCallable(segment_audio)
adjusted_segments = self.adjust_whisper_timestamp(segment_result["segments"], adjust_seconds=segment_start, max_source_time=segment_duration)
# Append to output
result['text'] += segment_result['text']
result['segments'].extend(adjusted_segments)
# Increment detected language
languageCounter[segment_result['language']] += 1
if len(languageCounter) > 0:
result['language'] = languageCounter.most_common(1)[0][0]
return result
def include_gaps(self, segments: Iterator[dict], min_gap_length: float, total_duration: float):
result = []
last_end_time = 0
for segment in segments:
segment_start = float(segment['start'])
segment_end = float(segment['end'])
if (last_end_time != segment_start):
delta = segment_start - last_end_time
if (min_gap_length is None or delta >= min_gap_length):
result.append( { 'start': last_end_time, 'end': segment_start, 'gap': True } )
last_end_time = segment_end
result.append(segment)
# Also include total duration if specified
if (total_duration is not None and last_end_time < total_duration):
delta = total_duration - segment_start
if (min_gap_length is None or delta >= min_gap_length):
result.append( { 'start': last_end_time, 'end': total_duration, 'gap': True } )
return result
# Expand the end time of each segment to the start of the next segment
def expand_gaps(self, segments: List[Dict[str, Any]], total_duration: float):
result = []
if len(segments) == 0:
return result
# Add gap at the beginning if needed
if (segments[0]['start'] > 0):
result.append({ 'start': 0, 'end': segments[0]['start'], 'gap': True } )
for i in range(len(segments) - 1):
current_segment = segments[i]
next_segment = segments[i + 1]
delta = next_segment['start'] - current_segment['end']
# Expand if the gap actually exists
if (delta >= 0):
current_segment = current_segment.copy()
current_segment['expand_amount'] = delta
current_segment['end'] = next_segment['start']
result.append(current_segment)
last_segment = result[-1]
# Also include total duration if specified
if (total_duration is not None):
last_segment = result[-1]
if (last_segment['end'] < total_duration):
last_segment = last_segment.copy()
last_segment['end'] = total_duration
result[-1] = last_segment
return result
def adjust_whisper_timestamp(self, segments: Iterator[dict], adjust_seconds: float, max_source_time: float = None):
result = []
for segment in segments:
segment_start = float(segment['start'])
segment_end = float(segment['end'])
# Filter segments?
if (max_source_time is not None):
if (segment_start > max_source_time):
continue
segment_end = min(max_source_time, segment_end)
new_segment = segment.copy()
# Add to start and end
new_segment['start'] = segment_start + adjust_seconds
new_segment['end'] = segment_end + adjust_seconds
result.append(new_segment)
return result
def pad_timestamps(self, timestamps: List[Dict[str, Any]], padding_left: float, padding_right: float):
if (padding_left == 0 and padding_right == 0):
return timestamps
result = []
for i in range(len(timestamps)):
prev_entry = timestamps[i - 1] if i > 0 else None
curr_entry = timestamps[i]
next_entry = timestamps[i + 1] if i < len(timestamps) - 1 else None
segment_start = curr_entry['start']
segment_end = curr_entry['end']
if padding_left is not None:
segment_start = max(prev_entry['end'] if prev_entry else 0, segment_start - padding_left)
if padding_right is not None:
segment_end = segment_end + padding_right
# Do not pad past the next segment
if (next_entry is not None):
segment_end = min(next_entry['start'], segment_end)
result.append({ 'start': segment_start, 'end': segment_end })
return result
def merge_timestamps(self, timestamps: List[Dict[str, Any]], max_merge_gap: float, max_merge_size: float):
if max_merge_gap is None:
return timestamps
result = []
current_entry = None
for entry in timestamps:
if current_entry is None:
current_entry = entry
continue
# Get distance to the previous entry
distance = entry['start'] - current_entry['end']
current_entry_size = current_entry['end'] - current_entry['start']
if distance <= max_merge_gap and (max_merge_size is None or current_entry_size <= max_merge_size):
# Merge
current_entry['end'] = entry['end']
else:
# Output current entry
result.append(current_entry)
current_entry = entry
# Add final entry
if current_entry is not None:
result.append(current_entry)
return result
def multiply_timestamps(self, timestamps: List[Dict[str, Any]], factor: float):
result = []
for entry in timestamps:
start = entry['start']
end = entry['end']
result.append({
'start': start * factor,
'end': end * factor
})
return result
class VadSileroTranscription(AbstractTranscription):
def __init__(self, segment_padding_left=SEGMENT_PADDING_LEFT, segment_padding_right=SEGMENT_PADDING_RIGHT,
max_silent_period=MAX_SILENT_PERIOD, max_merge_size=MAX_MERGE_SIZE, transcribe_non_speech: bool = False,
copy = None):
super().__init__(segment_padding_left=segment_padding_left, segment_padding_right=segment_padding_right,
max_silent_period=max_silent_period, max_merge_size=max_merge_size, transcribe_non_speech=transcribe_non_speech)
if copy:
self.model = copy.model
self.get_speech_timestamps = copy.get_speech_timestamps
else:
self.model, utils = torch.hub.load(repo_or_dir='snakers4/silero-vad', model='silero_vad')
(self.get_speech_timestamps, _, _, _, _) = utils
def get_transcribe_timestamps(self, audio: str):
wav = self.get_audio_segment(audio)
sample_timestamps = self.get_speech_timestamps(wav, self.model, sampling_rate=self.sampling_rate, threshold=SPEECH_TRESHOLD)
seconds_timestamps = self.multiply_timestamps(sample_timestamps, factor=1 / self.sampling_rate)
return seconds_timestamps
# A very simple VAD that just marks every N seconds as speech
class VadPeriodicTranscription(AbstractTranscription):
def __init__(self, periodic_duration: int):
super().__init__()
self.periodic_duration = periodic_duration
def get_transcribe_timestamps(self, audio: str):
# Get duration in seconds
audio_duration = float(ffmpeg.probe(audio)["format"]["duration"])
result = []
# Generate a timestamp every N seconds
start_timestamp = 0
while (start_timestamp < audio_duration):
end_timestamp = min(start_timestamp + self.periodic_duration, audio_duration)
segment_duration = end_timestamp - start_timestamp
# Minimum duration is 1 second
if (segment_duration >= 1):
result.append( { 'start': start_timestamp, 'end': end_timestamp } )
start_timestamp = end_timestamp
return result
def load_audio(file: str, sample_rate: int = 16000,
start_time: str = None, duration: str = None):
"""
Open an audio file and read as mono waveform, resampling as necessary
Parameters
----------
file: str
The audio file to open
sr: int
The sample rate to resample the audio if necessary
start_time: str
The start time, using the standard FFMPEG time duration syntax, or None to disable.
duration: str
The duration, using the standard FFMPEG time duration syntax, or None to disable.
Returns
-------
A NumPy array containing the audio waveform, in float32 dtype.
"""
try:
inputArgs = {'threads': 0}
if (start_time is not None):
inputArgs['ss'] = start_time
if (duration is not None):
inputArgs['t'] = duration
# This launches a subprocess to decode audio while down-mixing and resampling as necessary.
# Requires the ffmpeg CLI and `ffmpeg-python` package to be installed.
out, _ = (
ffmpeg.input(file, **inputArgs)
.output("-", format="s16le", acodec="pcm_s16le", ac=1, ar=sample_rate)
.run(cmd="ffmpeg", capture_stdout=True, capture_stderr=True)
)
except ffmpeg.Error as e:
raise RuntimeError(f"Failed to load audio: {e.stderr.decode()}")
return np.frombuffer(out, np.int16).flatten().astype(np.float32) / 32768.0