--- tags: - espnet - audio - diarization language: noinfo datasets: - mini_librispeech license: cc-by-4.0 --- ## ESPnet2 DIAR model ### `jkang/espnet2_mini_librispeech_diar` This model was trained by jaekookang using mini_librispeech recipe in [espnet](https://github.com/espnet/espnet/). ### Demo: How to use in ESPnet2 ```bash cd espnet git checkout e08a89e0a43db7fc12bec835c62a000ad10bd417 pip install -e . cd egs2/mini_librispeech/diar1 ./run.sh --skip_data_prep false --skip_train true --download_model jkang/espnet2_mini_librispeech_diar ``` # RESULTS ## Environments - date: `Tue Feb 8 16:41:16 KST 2022` - python version: `3.9.7 (default, Sep 16 2021, 13:09:58) [GCC 7.5.0]` - espnet version: `espnet 0.10.6a1` - pytorch version: `pytorch 1.10.1` - Git hash: `e08a89e0a43db7fc12bec835c62a000ad10bd417` - Commit date: `Sun Feb 6 18:54:20 2022 -0500` ## diar_train_diar_raw ### DER dev_clean_2_ns2_beta2_500 |threshold_median_collar|DER| |---|---| |result_th0.3_med11_collar0.0|31.39| |result_th0.3_med1_collar0.0|31.78| |result_th0.4_med11_collar0.0|29.99| |result_th0.4_med1_collar0.0|30.61| |result_th0.5_med11_collar0.0|29.28| |result_th0.5_med1_collar0.0|30.19| |result_th0.6_med11_collar0.0|29.50| |result_th0.6_med1_collar0.0|30.66| |result_th0.7_med11_collar0.0|30.90| |result_th0.7_med1_collar0.0|32.38| ## DIAR config
expand ``` config: conf/train_diar.yaml print_config: false log_level: INFO dry_run: false iterator_type: chunk output_dir: exp/diar_train_diar_raw 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: 100 patience: 3 val_scheduler_criterion: - valid - loss early_stopping_criterion: - valid - loss - min best_model_criterion: - - valid - acc - max keep_nbest_models: 3 nbest_averaging_interval: 0 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_matplotlib: true 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: attractor_weight: 1.0 use_preprocessor: true frontend: default frontend_conf: fs: 8k hop_length: 128 specaug: null specaug_conf: {} 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: {} attractor: null attractor_conf: {} required: - output_dir version: 0.10.6a1 distributed: false ```
### Citing ESPnet ```BibTex @inproceedings{watanabe2018espnet, author={Shinji Watanabe and Takaaki Hori and Shigeki Karita and Tomoki Hayashi and Jiro Nishitoba and Yuya Unno and Nelson Yalta and Jahn Heymann and Matthew Wiesner and Nanxin Chen and Adithya Renduchintala and Tsubasa Ochiai}, title={{ESPnet}: End-to-End Speech Processing Toolkit}, year={2018}, booktitle={Proceedings of Interspeech}, pages={2207--2211}, doi={10.21437/Interspeech.2018-1456}, url={http://dx.doi.org/10.21437/Interspeech.2018-1456} } ``` or arXiv: ```bibtex @misc{watanabe2018espnet, title={ESPnet: End-to-End Speech Processing Toolkit}, author={Shinji Watanabe and Takaaki Hori and Shigeki Karita and Tomoki Hayashi and Jiro Nishitoba and Yuya Unno and Nelson Yalta and Jahn Heymann and Matthew Wiesner and Nanxin Chen and Adithya Renduchintala and Tsubasa Ochiai}, year={2018}, eprint={1804.00015}, archivePrefix={arXiv}, primaryClass={cs.CL} } ```