21 lines
478 B
Markdown
21 lines
478 B
Markdown
|
# Routines for DeepSpeech features processing
|
||
|
|
Several routines for [DeepSpeech](https://github.com/mozilla/DeepSpeech) features processing, like speech features generation for [VOCA](https://github.com/TimoBolkart/voca) model.
|
||
|
|
|
||
|
|
## Installation
|
||
|
|
|
||
|
|
```
|
||
|
|
pip3 install -r requirements.txt
|
||
|
|
```
|
||
|
|
|
||
|
|
## Usage
|
||
|
|
|
||
|
|
Generate wav files:
|
||
|
|
```
|
||
|
|
python3 extract_wav.py --in-video=<you_data_dir>
|
||
|
|
```
|
||
|
|
|
||
|
|
Generate files with DeepSpeech features:
|
||
|
|
```
|
||
|
|
python3 extract_ds_features.py --input=<you_data_dir>
|
||
|
|
```
|