espnet/Bagpiper_TTS_SFT_Data_ESPnet

Bagpiper-TTS SFT Data in ESPnet format

This release converts espnet/Bagpiper_TTS_SFT_Data at revision f0eb112765d33f30ff6136a6fa5feaa84ca79880 into portable, lossless ESPnet dialogue bundles. Complete: all 738,123 source rows and 1,151 shards are converted and uploaded. RELEASE_VALIDATION.json records the full source/count/hash reconciliation.

Each .tar.zst contains original encoded audio, three-element dialogue messages, and source provenance. Audio is deduplicated by SHA-256 within each ordered conversion stream; identical files across streams are verified and merged during extraction. Repeated training examples remain present. All text, message roles, and order are preserved. The original globally unique row_id becomes ESPnet example_id. No audio resampling or lossy transcoding is performed.

Verified release counts

Partition	Preserved rows	Distinct audio files
classical_tts	235,279	235,077
general_purpose	135,614	135,309
intent_to_speech	153,568	153,281
multi_talker	64,659	63,351
role_play	47,116	47,108
svs	101,887	100,843

Distinct audio counts use (partition, SHA256) after reconciling all conversion streams. Archives may contain repeated files across streams; verified extraction merges them. Every original training row remains present.

VALIDATION.json covers 3,814 original-versus-converted sample rows across all eight partitions and 24 decoded waveform comparisons using the actual ESPnet reader. The full release additionally checks every input shard hash, every embedded audio hash, all row IDs, and every uploaded archive's size and LFS SHA256. GPU training and inference have not been validated. Use prepare_espnet.py for local paths and manifests; these bundles are not directly loadable with datasets.load_dataset.

Prepare local training inputs

python -m pip install huggingface_hub zstandard
hf download espnet/Bagpiper_TTS_SFT_Data_ESPnet prepare_espnet.py --local-dir tools
python tools/prepare_espnet.py --repo-id espnet/Bagpiper_TTS_SFT_Data_ESPnet --data-root /path/to/bagpiper-tts-sft

For a small prefix trial, add --max-shards 2. Repeat --partition NAME to select application partitions. The helper verifies archive SHA-256, extracts safely, resolves audio paths, checks duplicate IDs and linkage, and writes prepared/NAME/{all,train,valid}.json plus dialogues.jsonl.

The default validation fraction is 1%, chosen deterministically from the audio SHA-256. The same audio always has the same split, including across source shards and partitions. Use all.json to retain the original complete training pool. This suggested split is not the paper's official split. --valid-fraction 0 assigns every row to train.

From the ESPnet repository root, collect exact token lengths with the checkpoint-compatible training config, then pass the generated manifests to the recipe:

python -m espnet2.speechlm.bin.prepare_length_stats \
  --train-config /path/to/checkpoint-compatible.yaml \
  --train-unregistered-specifier 'dialogue:NAME_train:/path/to/prepared/NAME/train.json' \
  --valid-unregistered-specifier 'dialogue:NAME_valid:/path/to/prepared/NAME/valid.json' \
  --output-dir /path/to/stats

Use distinct dataset names for each partition and split. Do not select arbitrary archive subsets: later bundles can reference audio deduplicated into earlier bundles. The helper selects consecutive prefixes and handles this dependency.

Provenance and terms

SOURCE_MANIFEST.json pins every source shard and checksum. STATUS.json records conversion/upload checksums and counts. provenance/ retains original row IDs, source subsets/revisions, audio hashes and source annotations. CONVERSION.md and the executable converter document the release process.

This is a format conversion of the linked source dataset and does not grant new rights to its audio or annotations. Source terms and the source dataset's disclosed limitations apply. The general Bagpiper source release is the recipe-v1 pool; conversion does not make it the paper-v2 selected pool.

ESPnet recipes: Bagpiper · Bagpiper-TTS.