File size: 4,319 Bytes
2acac75 3ad3dbd 2acac75 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 |
---
tags:
- espnet
- audio
- diarization
language:
datasets:
- mini_librispeech
license: cc-by-4.0
---
## ESPnet2 DIAR model
### `YushiUeda/test`
This model was trained by Yushi Ueda using mini_librispeech recipe in [espnet](https://github.com/espnet/espnet/).
### Demo: How to use in ESPnet2
```bash
cd espnet
git checkout 4dfa2be4331d3d68f124aa5fd81f63217a7278a4
pip install -e .
cd egs2/mini_librispeech/diar1
./run.sh --skip_data_prep false --skip_train true --download_model YushiUeda/test
```
<!-- Generated by scripts/utils/show_diar_result.sh -->
# RESULTS
## Environments
- date: `Wed Aug 25 23:29:07 EDT 2021`
- python version: `3.7.11 (default, Jul 27 2021, 14:32:16) [GCC 7.5.0]`
- espnet version: `espnet 0.10.2a1`
- pytorch version: `pytorch 1.9.0+cu102`
- Git hash: `19bcd34f9395e01e54a97c4db5ecbcedb429dd92`
- Commit date: `Tue Aug 24 19:50:44 2021 -0400`
## `diar_train_diar_raw_max_epoch20`
### DER
`dev_clean_2_ns2_beta2_500`
|threshold_median_collar|DER|
|---|---|
|result_th0.3_med1_collar0.0|32.42|
|result_th0.3_med11_collar0.0|32.03|
|result_th0.4_med1_collar0.0|30.96|
|result_th0.4_med11_collar0.0|30.26|
|result_th0.5_med1_collar0.0|30.35|
|result_th0.5_med11_collar0.0|29.37|
|result_th0.6_med1_collar0.0|30.77|
|result_th0.6_med11_collar0.0|29.52|
|result_th0.7_med1_collar0.0|32.60|
|result_th0.7_med11_collar0.0|31.03|
## DIAR config
<details><summary>expand</summary>
```
config: conf/train_diar.yaml
print_config: false
log_level: INFO
dry_run: false
iterator_type: chunk
output_dir: exp/diar_train_diar_raw_max_epoch20
ngpu: 1
seed: 0
num_workers: 1
num_att_plot: 3
dist_backend: nccl
dist_init_method: env://
dist_world_size: null
dist_rank: null
local_rank: 0
dist_master_addr: null
dist_master_port: null
dist_launcher: null
multiprocessing_distributed: false
unused_parameters: false
sharded_ddp: false
cudnn_enabled: true
cudnn_benchmark: false
cudnn_deterministic: true
collect_stats: false
write_collected_feats: false
max_epoch: 20
patience: 3
val_scheduler_criterion:
- valid
- loss
early_stopping_criterion:
- valid
- loss
- min
best_model_criterion:
- - valid
- acc
- max
keep_nbest_models: 3
grad_clip: 5
grad_clip_type: 2.0
grad_noise: false
accum_grad: 2
no_forward_run: false
resume: true
train_dtype: float32
use_amp: false
log_interval: null
use_tensorboard: true
use_wandb: false
wandb_project: null
wandb_id: null
wandb_entity: null
wandb_name: null
wandb_model_log_interval: -1
detect_anomaly: false
pretrain_path: null
init_param: []
ignore_init_mismatch: false
freeze_param: []
num_iters_per_epoch: null
batch_size: 16
valid_batch_size: null
batch_bins: 1000000
valid_batch_bins: null
train_shape_file:
- exp/diar_stats_8k/train/speech_shape
- exp/diar_stats_8k/train/spk_labels_shape
valid_shape_file:
- exp/diar_stats_8k/valid/speech_shape
- exp/diar_stats_8k/valid/spk_labels_shape
batch_type: folded
valid_batch_type: null
fold_length:
- 80000
- 800
sort_in_batch: descending
sort_batch: descending
multiple_iterator: false
chunk_length: 200000
chunk_shift_ratio: 0.5
num_cache_chunks: 64
train_data_path_and_name_and_type:
- - dump/raw/simu/data/train_clean_5_ns2_beta2_500/wav.scp
- speech
- sound
- - dump/raw/simu/data/train_clean_5_ns2_beta2_500/espnet_rttm
- spk_labels
- rttm
valid_data_path_and_name_and_type:
- - dump/raw/simu/data/dev_clean_2_ns2_beta2_500/wav.scp
- speech
- sound
- - dump/raw/simu/data/dev_clean_2_ns2_beta2_500/espnet_rttm
- spk_labels
- rttm
allow_variable_data_keys: false
max_cache_size: 0.0
max_cache_fd: 32
valid_max_cache_size: null
optim: adam
optim_conf:
lr: 0.01
scheduler: noamlr
scheduler_conf:
warmup_steps: 1000
num_spk: 2
init: xavier_uniform
input_size: null
model_conf:
loss_type: pit
use_preprocessor: true
frontend: default
frontend_conf:
fs: 8k
hop_length: 128
normalize: global_mvn
normalize_conf:
stats_file: exp/diar_stats_8k/train/feats_stats.npz
encoder: transformer
encoder_conf:
input_layer: linear
num_blocks: 2
linear_units: 512
dropout_rate: 0.1
output_size: 256
attention_heads: 4
attention_dropout_rate: 0.0
decoder: linear
decoder_conf: {}
label_aggregator: label_aggregator
label_aggregator_conf: {}
required:
- output_dir
version: 0.10.2a1
distributed: false
```
</details>
|