Skip to content

Latest commit

 

History

History
58 lines (47 loc) · 2.19 KB

README.md

File metadata and controls

58 lines (47 loc) · 2.19 KB
Raw

tts_data_tools

Data processing tools for preparing speech and labels for training TTS voices

Installation

pip install git+https://github.com/zackhodari/tts_data_tools

Usage

Batch processing of datasets

tdt_process_dataset \
    --lab_dir DIR \
    --wav_dir DIR \
    --id_list FILE \
    --out_dir DIR \
    --question_file FILE \
    [--state_level | --no-state_level] \
    [--upsample_to_frame_level] \
    [--trim_silences] \
    [--subphone_feat_type STR] \
    [--calculate_normalisation] \
    [--normalisation_of_deltas]

Other batch processing scripts are defined in setup.py. Additional example scripts are given in scripts/, these can be used from the command line or as templates for your own pre-processing.

Label creation

lab_gen provides four tools to take text and convert them to numerical label features ready for training.

txt_to_utt.py and utt_to_lab.py wrap Festival (to be installed separately), and can extract full-context HTS-style labels.

align_lab.py wraps HTK (to be installed separately) and performs forced alignment using the wavform.

If you already have forced-aligned labels you can use lab_to_feat.py to convert the text-formatted features to numpy arrays (saved as .npy):

tdt_lab_to_feat \
    --lab_dir DIR \
    --id_list FILE \
    --out_dir DIR \
    --question_file FILE \
    [--state_level | --no-state_level] \
    [--upsample_to_frame_level] \
    [--subphone_feat_type STR] \
    [--calculate_normalisation]

Acoustic feature extraction

wav_gen provides a few common vocoders that can be used to extract typical TTS acoustic features. This includes F0 extraction with REAPER and full vocoder feature extraction with WORLD, which can be used as follows.

tdt_world \
    --wav_dir DIR \
    --id_list FILE \
    --out_dir DIR \
    [--calculate_normalisation] \
    [--normalisation_of_deltas]