NagaYu/bleep-spans

Bleep spans — synthetic sensitive-speech regions with frame-accurate labels

Where sensitive information is spoken, and what kind it is — never what was said.

Every recording is synthetic. No real telephone call, clinical recording, or any other real speech was used, recorded, or derived from at any stage.

• 🤗 Model: NagaYu/bleep-0.09b
• 🎛️ Demo: NagaYu/bleep

What a row contains

utt_id, voice_key, condition, duration, subsets, and three parallel arrays — span_starts, span_ends, span_labels.

There is no transcript field, and no field containing what was said. The corpus annotates where sensitive speech occurs and what kind it is. That is a deliberate limit: a public dataset of spoken account numbers, even invented ones, is a worse artefact than one that only marks their positions.

Two configs, because of a licensing split

config	speech	audio included?	why
features	macOS system voices — realistic, 24 English speakers	❌ 64-band log-mel (fp16) + annotations	Apple's system voices are licensed for use, not for republication as a corpus. Features are what the model consumes anyway, and are not invertible to intelligible speech in any practical sense (magnitude only, no phase, 64 bands).
audio	Bleep's own numpy source-filter synthesiser	✅ full 16 kHz waveforms	We own the output outright, so it is freely redistributable. The speech is pseudo-speech: correct syllable rhythm, formant structure and prosody, but not intelligible words.

The published model was trained on features. The audio config is the same generator, conditions, splits and seed, rendered with the synthesiser we own — so you can listen to it, re-derive features, or use it to test a pipeline end to end.

from datasets import load_dataset

feats = load_dataset("NagaYu/bleep-spans", "features", split="test")
audio = load_dataset("NagaYu/bleep-spans", "audio", split="test")

Splits

Three speaker-disjoint partitions — no voice appears in more than one — plus a fourth evaluation set that reuses the test speakers with degradation conditions withheld from training, so a drop there is attributable to the condition rather than the voice.

calib exists because a threshold chosen on the test set makes "recall at a fixed false-alarm rate" a fitted parameter rather than a measurement — and choosing it on the training speakers would be optimistic in a subtler way, since the score distribution of a seen speaker is not that of an unseen one.

features config (13.15 hours)

split	utterances	hours	spans	voices	negatives
train	2200	9.15	2828	13	591
calib	250	0.94	278	4	37
test	550	1.99	641	7	77
test_heldout	300	1.07	345	7	39

audio config (14.64 hours)

split	utterances	hours	spans	voices	negatives
train	2200	10.16	2796	13	580
calib	250	1.02	288	4	32
test	550	2.26	629	7	67
test_heldout	300	1.21	345	7	38

Categories

• DIGIT_SEQUENCE
• GROUPED_DIGITS
• SPELLED_OUT
• DATETIME
• ADDRESS
• PERSON_NAME
• IDENTIFIER
• MONEY

Roughly balanced; on the test split: DIGIT_SEQUENCE 79, GROUPED_DIGITS 87, SPELLED_OUT 96, DATETIME 94, ADDRESS 63, PERSON_NAME 76, IDENTIFIER 85, MONEY 61.

The four weighted most heavily — GROUPED_DIGITS, SPELLED_OUT, DATETIME, IDENTIFIER — are the reported failure modes of transcript-based detection: digits read in chunks, names spelled letter by letter, relative time expressions, and identifiers spoken as separate elements. Utterances carry subsets tags (hard:grouped_digits, hard:spelled, hard:overlap, …) so these can be sliced directly.

How the labels are exact

Each utterance is rendered segment by segment — carrier speech, then the sensitive item, then more carrier — and concatenated. A span is the sample range the sensitive segment occupies by construction. There is no forced aligner in the loop and therefore no alignment error in the labels.

Degradation conditions

All strictly time-preserving, so the labels stay valid. Anything that stretched or delayed the signal would silently desynchronise every span, and nothing downstream would notice.

In training (8): clean, telephony_8k, noisy_babble_10db, noisy_street_5db, reverb_moderate, mic_variation, overlap_6db, telephony_noisy

Held out (5): packet_loss_voip, wind_noise_3db, reverb_heavy_phone, overlap_0db_three_way, telephony_overlap_3db

Two controls that protect a benchmark built on this corpus

Announcing-carrier decorrelation. Sensitive items are usually announced ("the account number is …"). If every announcement were followed by a sensitive item, a model could score well by detecting the announcement and never modelling the sensitive region — and a benchmark would report that as success. 35% of announcing carriers are followed by a harmless continuation instead.

Join-artefact decorrelation. Concatenating separately-rendered segments makes every segment boundary a potential artefact. If joins only ever occurred beside sensitive material, "a join is here" would itself predict sensitivity. Carrier phrases are therefore split into separately-rendered clauses too, so joins appear everywhere, including inside negative utterances. Measured ratio of non-sensitive to sensitive-adjacent joins: 1.01.

Known limitations

It is synthetic. Synthetic speech has narrower prosodic and disfluency variety than real conversation: no false starts, no laughter, no emotional range. Recall measured on it is an estimate of recall on real audio, not a measurement of it.

The degradations turned out to be too mild. In the accompanying benchmark, a whisper-tiny.en + text-PII baseline scores 0.818 on clean audio and 0.857 at 8 kHz — i.e. better. A band-limit plus G.711 is not a hard condition for a modern ASR on hyper-articulated TTS speech. The baseline only collapses in the held-out set (0.271 at 0 dB three-way overlap). If you are using this corpus to study transcription failure, start from the held-out conditions, and consider making them harsher still.

Concatenation removes coarticulation. Utterances are rendered segment-by-segment and concatenated, so there is no coarticulation across a carrier/sensitive boundary. Mitigations: randomised 30-160 ms inter-segment gaps, a 15 ms crossfade at each join, and the fact that every evaluation condition applies reverberation, noise and/or codec processing after concatenation, which smears join artefacts. A residual concern remains that the tagger could key on join artefacts rather than on speech; test_synth.py::test_join_artifacts_are_not_a_cue checks that carrier/carrier joins (present in every utterance, including negatives) are as frequent as carrier/sensitive joins, so a join is not by itself evidence of sensitivity.

English only. Japanese generators exist in the code and are wired end to end, but are not represented here and are not benchmarked.

Provenance

Generated by scripts/build_dataset.py (seed 20260918). The build is deterministic and resumable. Names, addresses and numbers are invented from word lists; any collision with a real person is coincidental and carries no information, because the strings exist only to drive a speech synthesiser and are discarded before any artefact is written.

Licence

Apache-2.0 for the annotations, the audio config, and the generator code.