Skip silence around hallucinations (#1838)

* Add clip_timestamps option * Add hallucination_silence_threshold option * Fix typing for python < 3.9 --------- Co-authored-by: Jong Wook Kim <[email protected]>
openai · Dec 18, 2023 · ba3f3cd · ba3f3cd
1 parent 8bc8860
commit ba3f3cd
Show file tree

Hide file tree

Showing 3 changed files with 153 additions and 19 deletions.
diff --git a/whisper/timing.py b/whisper/timing.py
@@ -299,6 +299,7 @@ def add_word_timestamps(
     word_durations = np.array([t.end - t.start for t in alignment])
     word_durations = word_durations[word_durations.nonzero()]
     median_duration = np.median(word_durations) if len(word_durations) > 0 else 0.0
+    median_duration = min(0.7, float(median_duration))
     max_duration = median_duration * 2
 
     # hack: truncate long words at sentence boundaries.

diff --git a/whisper/transcribe.py b/whisper/transcribe.py
@@ -2,7 +2,7 @@
 import os
 import traceback
 import warnings
-from typing import TYPE_CHECKING, Optional, Tuple, Union
+from typing import TYPE_CHECKING, List, Optional, Tuple, Union
 
 import numpy as np
 import torch
@@ -23,6 +23,7 @@
 from .utils import (
     exact_div,
     format_timestamp,
+    get_end,
     get_writer,
     make_safe,
     optional_float,
@@ -48,6 +49,8 @@ def transcribe(
     word_timestamps: bool = False,
     prepend_punctuations: str = "\"'“¿([{-",
     append_punctuations: str = "\"'.。,，!！?？:：”)]}、",
+    clip_timestamps: Union[str, List[float]] = "0",
+    hallucination_silence_threshold: Optional[float] = None,
     **decode_options,
 ):
     """
@@ -102,6 +105,14 @@ def transcribe(
     decode_options: dict
         Keyword arguments to construct `DecodingOptions` instances
 
+    clip_timestamps: Union[str, List[float]]
+        Comma-separated list start,end,start,end,... timestamps (in seconds) of clips to process.
+        The last end timestamp defaults to the end of the file.
+
+    hallucination_silence_threshold: Optional[float]
+        When word_timestamps is True, skip silent periods longer than this threshold (in seconds)
+        when a possible hallucination is detected
+
     Returns
     -------
     A dictionary containing the resulting text ("text") and segment-level details ("segments"), and
@@ -121,6 +132,7 @@ def transcribe(
     # Pad 30-seconds of silence to the input audio, for slicing
     mel = log_mel_spectrogram(audio, model.dims.n_mels, padding=N_SAMPLES)
     content_frames = mel.shape[-1] - N_FRAMES
+    content_duration = float(content_frames * HOP_LENGTH / SAMPLE_RATE)
 
     if decode_options.get("language", None) is None:
         if not model.is_multilingual:
@@ -147,6 +159,19 @@ def transcribe(
         task=task,
     )
 
+    if isinstance(clip_timestamps, str):
+        clip_timestamps = [
+            float(ts) for ts in (clip_timestamps.split(",") if clip_timestamps else [])
+        ]
+    seek_points: List[int] = [round(ts * FRAMES_PER_SECOND) for ts in clip_timestamps]
+    if len(seek_points) == 0:
+        seek_points.append(0)
+    if len(seek_points) % 2 == 1:
+        seek_points.append(content_frames)
+    seek_clips: List[Tuple[int, int]] = list(zip(seek_points[::2], seek_points[1::2]))
+
+    punctuation = "\"'“¿([{-\"'.。,，!！?？:：”)]}、"
+
     if word_timestamps and task == "translate":
         warnings.warn("Word-level timestamps on translations may not be reliable.")
 
@@ -190,7 +215,8 @@ def decode_with_fallback(segment: torch.Tensor) -> DecodingResult:
 
         return decode_result
 
-    seek = 0
+    clip_idx = 0
+    seek = seek_clips[clip_idx][0]
     input_stride = exact_div(
         N_FRAMES, model.dims.n_audio_ctx
     )  # mel frames per output token: 2
@@ -229,10 +255,23 @@ def new_segment(
         total=content_frames, unit="frames", disable=verbose is not False
     ) as pbar:
         last_speech_timestamp = 0.0
-        while seek < content_frames:
+        # NOTE: This loop is obscurely flattened to make the diff readable.
+        # A later commit should turn this into a simpler nested loop.
+        # for seek_clip_start, seek_clip_end in seek_clips:
+        #     while seek < seek_clip_end
+        while clip_idx < len(seek_clips):
+            seek_clip_start, seek_clip_end = seek_clips[clip_idx]
+            if seek < seek_clip_start:
+                seek = seek_clip_start
+            if seek >= seek_clip_end:
+                clip_idx += 1
+                if clip_idx < len(seek_clips):
+                    seek = seek_clips[clip_idx][0]
+                continue
             time_offset = float(seek * HOP_LENGTH / SAMPLE_RATE)
-            mel_segment = mel[:, seek : seek + N_FRAMES]
-            segment_size = min(N_FRAMES, content_frames - seek)
+            window_end_time = float((seek + N_FRAMES) * HOP_LENGTH / SAMPLE_RATE)
+            segment_size = min(N_FRAMES, content_frames - seek, seek_clip_end - seek)
+            mel_segment = mel[:, seek : seek + segment_size]
             segment_duration = segment_size * HOP_LENGTH / SAMPLE_RATE
             mel_segment = pad_or_trim(mel_segment, N_FRAMES).to(model.device).to(dtype)
 
@@ -257,6 +296,30 @@ def new_segment(
             previous_seek = seek
             current_segments = []
 
+            # anomalous words are very long/short/improbable
+            def word_anomaly_score(word: dict) -> float:
+                probability = word.get("probability", 0.0)
+                duration = word["end"] - word["start"]
+                score = 0.0
+                if probability < 0.15:
+                    score += 1.0
+                if duration < 0.133:
+                    score += (0.133 - duration) * 15
+                if duration > 2.0:
+                    score += duration - 2.0
+                return score
+
+            def is_segment_anomaly(segment: Optional[dict]) -> bool:
+                if segment is None or not segment["words"]:
+                    return False
+                words = [w for w in segment["words"] if w["word"] not in punctuation]
+                words = words[:8]
+                score = sum(word_anomaly_score(w) for w in words)
+                return score >= 3 or score + 0.01 >= len(words)
+
+            def next_words_segment(segments: List[dict]) -> Optional[dict]:
+                return next((s for s in segments if s["words"]), None)
+
             timestamp_tokens: torch.Tensor = tokens.ge(tokenizer.timestamp_begin)
             single_timestamp_ending = timestamp_tokens[-2:].tolist() == [False, True]
 
@@ -330,17 +393,71 @@ def new_segment(
                     append_punctuations=append_punctuations,
                     last_speech_timestamp=last_speech_timestamp,
                 )
-                word_end_timestamps = [
-                    w["end"] for s in current_segments for w in s["words"]
-                ]
-                if len(word_end_timestamps) > 0:
-                    last_speech_timestamp = word_end_timestamps[-1]
-                if not single_timestamp_ending and len(word_end_timestamps) > 0:
-                    seek_shift = round(
-                        (word_end_timestamps[-1] - time_offset) * FRAMES_PER_SECOND
-                    )
-                    if seek_shift > 0:
-                        seek = previous_seek + seek_shift
+
+                if not single_timestamp_ending:
+                    last_word_end = get_end(current_segments)
+                    if last_word_end is not None and last_word_end > time_offset:
+                        seek = round(last_word_end * FRAMES_PER_SECOND)
+
+                # skip silence before possible hallucinations
+                if hallucination_silence_threshold is not None:
+                    threshold = hallucination_silence_threshold
+                    if not single_timestamp_ending:
+                        last_word_end = get_end(current_segments)
+                        if last_word_end is not None and last_word_end > time_offset:
+                            remaining_duration = window_end_time - last_word_end
+                            if remaining_duration > threshold:
+                                seek = round(last_word_end * FRAMES_PER_SECOND)
+                            else:
+                                seek = previous_seek + segment_size
+
+                    # if first segment might be a hallucination, skip leading silence
+                    first_segment = next_words_segment(current_segments)
+                    if first_segment is not None and is_segment_anomaly(first_segment):
+                        gap = first_segment["start"] - time_offset
+                        if gap > threshold:
+                            seek = previous_seek + round(gap * FRAMES_PER_SECOND)
+                            continue
+
+                    # skip silence before any possible hallucination that is surrounded
+                    # by silence or more hallucinations
+                    hal_last_end = last_speech_timestamp
+                    for si in range(len(current_segments)):
+                        segment = current_segments[si]
+                        if not segment["words"]:
+                            continue
+                        if is_segment_anomaly(segment):
+                            next_segment = next_words_segment(
+                                current_segments[si + 1 :]
+                            )
+                            if next_segment is not None:
+                                hal_next_start = next_segment["words"][0]["start"]
+                            else:
+                                hal_next_start = time_offset + segment_duration
+                            silence_before = (
+                                segment["start"] - hal_last_end > threshold
+                                or segment["start"] < threshold
+                                or segment["start"] - time_offset < 2.0
+                            )
+                            silence_after = (
+                                hal_next_start - segment["end"] > threshold
+                                or is_segment_anomaly(next_segment)
+                                or window_end_time - segment["end"] < 2.0
+                            )
+                            if silence_before and silence_after:
+                                seek = round(
+                                    max(time_offset + 1, segment["start"])
+                                    * FRAMES_PER_SECOND
+                                )
+                                if content_duration - segment["end"] < threshold:
+                                    seek = content_frames
+                                current_segments[si:] = []
+                                break
+                        hal_last_end = segment["end"]
+
+                last_word_end = get_end(current_segments)
+                if last_word_end is not None:
+                    last_speech_timestamp = last_word_end
 
             if verbose:
                 for segment in current_segments:
@@ -427,6 +544,8 @@ def valid_model_name(name):
     parser.add_argument("--max_line_count", type=optional_int, default=None, help="(requires --word_timestamps True) the maximum number of lines in a segment")
     parser.add_argument("--max_words_per_line", type=optional_int, default=None, help="(requires --word_timestamps True, no effect with --max_line_width) the maximum number of words in a segment")
     parser.add_argument("--threads", type=optional_int, default=0, help="number of threads used by torch for CPU inference; supercedes MKL_NUM_THREADS/OMP_NUM_THREADS")
+    parser.add_argument("--clip_timestamps", type=str, default="0", help="comma-separated list start,end,start,end,... timestamps (in seconds) of clips to process, where the last end timestamp defaults to the end of the file")
+    parser.add_argument("--hallucination_silence_threshold", type=optional_float, help="(requires --word_timestamps True) skip silent periods longer than this threshold (in seconds) when a possible hallucination is detected")
     # fmt: on
 
     args = parser.parse_args().__dict__

diff --git a/whisper/utils.py b/whisper/utils.py
@@ -3,7 +3,7 @@
 import re
 import sys
 import zlib
-from typing import Callable, Optional, TextIO
+from typing import Callable, List, Optional, TextIO
 
 system_encoding = sys.getdefaultencoding()
 
@@ -68,6 +68,20 @@ def format_timestamp(
     )
 
 
+def get_start(segments: List[dict]) -> Optional[float]:
+    return next(
+        (w["start"] for s in segments for w in s["words"]),
+        segments[0]["start"] if segments else None,
+    )
+
+
+def get_end(segments: List[dict]) -> Optional[float]:
+    return next(
+        (w["end"] for s in reversed(segments) for w in reversed(s["words"])),
+        segments[-1]["end"] if segments else None,
+    )
+
+
 class ResultWriter:
     extension: str
 
@@ -129,8 +143,8 @@ def iterate_subtitles():
             line_len = 0
             line_count = 1
             # the next subtitle to yield (a list of word timings with whitespace)
-            subtitle: list[dict] = []
-            last = result["segments"][0]["words"][0]["start"]
+            subtitle: List[dict] = []
+            last: float = get_start(result["segments"]) or 0.0
             for segment in result["segments"]:
                 chunk_index = 0
                 words_count = max_words_per_line