Defects Missed in Transcription — AI Speaks After 0.5-Second Silence

A developer at forge.workstyle.tech discovered that speech-to-text (STT) based quality control for text-to-speech (TTS) models can miss defects where the model produces sounds not in the script after a 0.5-second silence. The developer found that Whisper STT dropped these short utterances, leading to a false 100% pass rate, and developed a waveform-based detection method using RMS envelope segmentation to catch such artifacts.

📝 Originally published in Japanese at forge.workstyle.tech . I used to perform quality control QC for TTS models using this process: I created 12 voices and passed all of them through this QC. Whisper got 4/4 accuracy, no trailing elongation, and sound pressure was within normal range. I reported 100% pass rate . Later, when I rechecked from a different angle, 4 of them still had defects . These were invisible to STT-based inspection due to fundamental limitations. The first clue came when I received this report: ご覧ください。 Total: 1.61s Body: 0.88s → Silence: 0.48s → 【0.16s utterance】 こちらです。 Total: 1.65s Body: 0.72s → Silence: 0.56s → 【0.28s utterance】 After finishing the script, there’s a full 0.5-second silence followed by a 0.1–0.3 second utterance. This isn’t trailing resonance— the model is producing sounds not in the script the root cause was training corpus contamination: "3 characters" allowed by the quality gate became verbal tics https://forge.workstyle.tech/blog/three-chars-became-a-verbal-tic/ . The reason my initial inspection missed this is simple: Whisper dropped these sounds . ご覧ください。 → STT: "ご覧くださいああ" ← barely caught こちらです。 → STT: "こちらです" ← completely dropped A 0.28-second utterance doesn’t appear in the transcription at all. Short sounds that aren’t meaningful words may not appear in STT output. As long as you’re only looking at transcriptions, this defect doesn’t exist. I even concluded, “STT got 0/6, so no extra sounds,” mistaking the blind spot of my measurement method for a property of the target. What’s actually being output appears in the waveform. By extracting voiced blocks from the RMS envelope and examining their sequence, we can detect these artifacts. python def segments wav bytes, thr ratio=0.06 : """Returns start sec, end sec , ... of voiced blocks""" w = wave.open io.BytesIO wav bytes ; sr = w.getframerate x = np.frombuffer w.readframes w.getnframes , dtype=np.int16 / 32768 W, H = int sr 0.020 , int sr 0.010 20ms window / 10ms hop rms = np.array np.sqrt np.mean x i H:i H+W 2 for i in range max 0, len x -W //H Use the larger of relative or absolute threshold act = rms max rms.max thr ratio, 0.004 segs, s = , None for i, a in enumerate act : if a and s is None: s = i elif not a and s is not None: if i - s 0.010 = 0.03: Ignore blocks <30ms segs.append s 0.010, i 0.010 s = None if s is not None: segs.append s 0.010, len act 0.010 return segs The max rms.max 0.06, 0.004 part is subtly critical. Relative threshold alone fails for low-volume voices. If the overall volume is quiet, the maximum value is small, causing noise floor to be misclassified as voiced. Absolute threshold alone fails for high-volume voices. Breathing or lip smacks get classified as voiced. The 12 voices had sound pressure ranging from −13.3 to −18.8 dB, so neither threshold alone could work across all voices. Discarding blocks under 30ms is also necessary. Without this, lip noise or quantization noise appears as many tiny blocks, breaking downstream logic. Once voiced blocks are extracted, we check: “Is there sufficient silence before the final block, and does that block have sufficient duration?” GAP MIN = 0.25 Silence this long or more indicates a separate utterance TAIL MIN = 0.06 Duration this long or more indicates an artifact def has trailing artifact wav : segs = segments wav if len segs < 2: return None No artifact if only one block gap = segs -1 0 - segs -2 1 Silence before last block tail = segs -1 1 - segs -1 0 Duration of last block if gap = GAP MIN and tail = TAIL MIN: return gap, tail return None GAP MIN=0.25 separates natural trailing resonance or pauses from clearly separated utterances. Measured artifacts had silence gaps of 0.26–0.91 seconds, so 0.25 is sufficient. TAIL MIN=0.06 avoids catching fade-out tails. Measured artifacts were 0.07–0.36 seconds long. In my first scan, all 12 models triggered the detector . The probe sentence contained this: では、始めます。 Block 0 = "では" Block 1 = "始めます" ← 0.40s silence followed by 0.75s duration This was a pause after a comma. After “では、” there’s a gap, then “始めます。” follows. The final block is part of the script itself, yet it perfectly matches the detection condition gap + subsequent utterance . The condition “there’s utterance after the final gap” will always produce false positives for sentences containing commas. That’s because it doesn’t consider script structure. There are two fixes: Limit probes to single sentences. If you exclude sentences with commas, any utterance after the body can be definitively identified as an artifact. This is what I adopted—simple implementation and no dependency on the script. Align with script end position. Derive the script end position from Whisper segments and check if energy exists beyond that point. This is more general but reintroduces STT dependency. If artifacts don’t appear in Whisper segments, the end position might be incorrectly determined. After removing false positives: Before with commas : 29 / 72 detections ← all 12 models triggered After single sentences only : 16 / 72 detections ← only 4 models triggered | Model | Artifacts | |---|---| | Male Narrator | 6/6 | | Female Operator | 4/6 | | Female Presenter | 4/6 | | Male Presenter | 2/6 | | Remaining 8 | 0/6 | Had I reported the initial results as-is, I would have spread false panic of “all 12 failed.” Once you build a detector, you must first test it on things that should not trigger it . What I learned is that audio quality inspection requires multiple methods that reveal different layers : | Method | Reveals | Misses | |---|---|---| | STT transcription | Word omissions, substitutions, large insertions | Short artifacts, silence structure, audio quality | | Waveform envelope | Utterance boundaries, silence, artifacts | What it’s saying | | Acoustic features F0, sound pressure, intonation | Pitch, volume, variation | Correctness of content | | Listening test | Everything but subjective & not scalable | — | Relying only on STT for QC meant assuming everything visible in the most familiar tool would be visible everywhere . In reality, STT only shows “what can be recognized as words.” Interestingly, these 4-second artifacts are hard to notice even when listening . A 0.1-second sound feels like “some lingering resonance” unless you’re paying close attention. Only by laying out the numbers do you realize there’s an abnormal structure: “silence 0.5s followed by sound.” If humans can’t perceive it subjectively, machines must measure it. And every measurement method has its own blind spots. Reflecting on this experience, here’s the process I should have followed: The fourth point is the key lesson: when detection rates are too high, suspect the detector, not the targets . A record of designing voices from single captions, manufacturing training corpora, and mass-producing role-specific practical voices. This article is Part 3: Quality Gate . ← Previous: Weeding out candidates using fixable defects https://forge.workstyle.tech/blog/measuring-factory-defects-as-product-traits/ → Next: How 70 minutes of training material vanished in an instant due to a network blink Full series 18 parts The insights in this article are compiled in the Diffusion TTS Manufacturing Pipeline for Mass-producing Practical Voices https://forge.workstyle.tech/blog/diffusion-tts-manufacturing-pipeline/ .