# speaker_code

**Repository Path**: zhitiankai001/speaker_code

## Basic Information

- **Project Name**: speaker_code
- **Description**: No description available
- **Primary Language**: Unknown
- **License**: Not specified
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 2
- **Created**: 2021-09-14
- **Last Updated**: 2021-09-14

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# Speaker Trainer toolkit

### How to Apply Webrtc Voice Activity Detection(VAD) in Your Own Dataset

`scripts/vad.py` will find the all `*.wav` in `--data_dir` and store VAD results to `*.vad`.

```bash
# usage
python3 scripts/vad.py --data_dir ${your_dataset_dir_path} --num_jobs 30
```

And then, build the train list and format the trials via these commands:

```bash
# usage
python3 scripts/build_datalist.py \
		--extension vad \
		--dataset_dir ${your_dataset_dir_path} \
		--data_list_path data_list.csv
```

### Extract Speaker Embeddings For Other Tasks

```bash
python3 extract_speaker_embedding.py \
	--nnet_type resnet34_quarter \
	--pooling_type ASP \
	--embedding_dim 256 \
	--checkpoint_path resnet34_quarter_sap_256.ckpt \
	--wave_path test.wav \
	--device cpu
```