VCTK
VCTK is an open English speech corpus. We provide examples for building Transformer models on this dataset.
Data preparation
Download data, create splits and generate audio manifests with
python -m examples.speech_synthesis.preprocessing.get_vctk_audio_manifest \
--output-data-root ${AUDIO_DATA_ROOT} \
--output-manifest-root ${AUDIO_MANIFEST_ROOT}
To denoise audio and trim leading/trailing silence using signal processing based VAD, run
for SPLIT in dev test train; do
python -m examples.speech_synthesis.preprocessing.denoise_and_vad_audio \
--audio-manifest ${AUDIO_MANIFEST_ROOT}/${SPLIT}.audio.tsv \
--output-dir ${PROCESSED_DATA_ROOT} \
--denoise --vad --vad-agg-level 3
done
which generates a new audio TSV manifest under ${PROCESSED_DATA_ROOT} with updated path to the processed audio and
a new column for SNR.
To do filtering by CER, follow the Automatic Evaluation section to
run ASR model (add --eval-target to get_eval_manifest for evaluation on the reference audio; add --err-unit char
to eval_asr to compute CER instead of WER). The example-level CER is saved to
${EVAL_OUTPUT_ROOT}/uer_cer.${SPLIT}.tsv.
Then, extract log-Mel spectrograms, generate feature manifest and create data configuration YAML with
python -m examples.speech_synthesis.preprocessing.get_feature_manifest \
--audio-manifest-root ${PROCESSED_DATA_ROOT} \
--output-root ${FEATURE_MANIFEST_ROOT} \
--ipa-vocab --use-g2p \
--snr-threshold 15 \
--cer-threshold 0.1 --cer-tsv-path ${EVAL_OUTPUT_ROOT}/uer_cer.${SPLIT}.tsv
where we use phoneme inputs (--ipa-vocab --use-g2p) as example. For sample filtering, we set the SNR and CER threshold
to 15 and 10%, respectively.
Training
(Please refer to the LJSpeech example.)
Inference
(Please refer to the LJSpeech example.)
Automatic Evaluation
(Please refer to the LJSpeech example.)
Results
| --arch | Params | Test MCD | Model |
|---|---|---|---|
| tts_transformer | 54M | 3.4 | Download |