{
  "models": {
    "tts": [
      "facebook/mms-tts-eng",
      "c71de0fe7204c83f1c10820a7d696d0b450048ba"
    ],
    "asr": [
      "openai/whisper-tiny.en",
      "87c7102498dcde7456f24cfd30239ca606ed9063"
    ]
  },
  "texts": [
    "A blue cup is on the table.",
    "The workshop opens at nine and closes at six.",
    "Please pick up order number three seven two.",
    "There are two red circles and one blue square."
  ],
  "source_sha256": "baa68968b393d97b2212740d79a68942259b3bfce2a0280994dc4829b1aa5b12",
  "tts": [
    {
      "id": 0,
      "text": "A blue cup is on the table.",
      "seed": 2026,
      "seconds": 0.5919178328476846,
      "samples": 35584,
      "pcm_gain": 1.0,
      "raw_sha256": "4b8b4a35fcc012d0a61e060cdeb17ca1c44ee131da3fecc0aacdbeaf1f27b59b"
    },
    {
      "id": 1,
      "text": "The workshop opens at nine and closes at six.",
      "seed": 2027,
      "seconds": 0.5662717088125646,
      "samples": 47104,
      "pcm_gain": 1.0,
      "raw_sha256": "3c278a7f225290ec14898f28476eba3b8d674123a1f62953ab2b4ce72fd3ff3a"
    },
    {
      "id": 2,
      "text": "Please pick up order number three seven two.",
      "seed": 2028,
      "seconds": 0.34792683320119977,
      "samples": 46592,
      "pcm_gain": 1.0,
      "raw_sha256": "e4d9474d4166e93331bb41ad56e2ccf7ab5001ae11cec03185a77e940d26ad62"
    },
    {
      "id": 3,
      "text": "There are two red circles and one blue square.",
      "seed": 2029,
      "seconds": 0.37532208394259214,
      "samples": 56064,
      "pcm_gain": 1.0,
      "raw_sha256": "2149de3c2fda19c60cda5e167f140a6bc6817c5f0535d772ba4e22e264335876"
    }
  ],
  "asr": [
    {
      "id": "0-clean",
      "condition": "clean",
      "audio": "0-clean.wav",
      "reference": "A blue cup is on the table.",
      "snr_db": null,
      "mix_gain": 1.0,
      "samples": 35584,
      "duration": 2.224,
      "audio_sha256": "ffa6638267a02cce202dd1451ead3293bcd8243483c598e0a3a37dfe3de53ef9",
      "feature_shape": [
        1,
        80,
        3000
      ],
      "token_ids": [
        50257,
        50362,
        317,
        4171,
        6508,
        318,
        319,
        262,
        3084,
        13,
        50256
      ],
      "text": " A blue cup is on the table.",
      "eos": true,
      "seconds": 0.8832078748382628,
      "rtf": 0.39712584300281595,
      "score": {
        "substitutions": 0,
        "deletions": 0,
        "insertions": 0,
        "reference_words": 7,
        "errors": 0,
        "wer": 0.0
      }
    },
    {
      "id": "0-snr20",
      "condition": "snr20",
      "audio": "0-snr20.wav",
      "reference": "A blue cup is on the table.",
      "snr_db": 20,
      "mix_gain": 1.0,
      "samples": 35584,
      "duration": 2.224,
      "audio_sha256": "debc2560e7ecd4822c13b134d5ca2e411ea737e454a5efe1fd11ed9431b0ac6f",
      "feature_shape": [
        1,
        80,
        3000
      ],
      "token_ids": [
        50257,
        50362,
        317,
        4171,
        6508,
        318,
        319,
        262,
        3084,
        13,
        50256
      ],
      "text": " A blue cup is on the table.",
      "eos": true,
      "seconds": 1.8140621669590473,
      "rtf": 0.8156754347837443,
      "score": {
        "substitutions": 0,
        "deletions": 0,
        "insertions": 0,
        "reference_words": 7,
        "errors": 0,
        "wer": 0.0
      }
    },
    {
      "id": "0-snr0",
      "condition": "snr0",
      "audio": "0-snr0.wav",
      "reference": "A blue cup is on the table.",
      "snr_db": 0,
      "mix_gain": 0.8821703281314965,
      "samples": 35584,
      "duration": 2.224,
      "audio_sha256": "4ff7095b9d94f371bda326dd82815830b499befa345449ff6ffc369fd81f7a75",
      "feature_shape": [
        1,
        80,
        3000
      ],
      "token_ids": [
        50257,
        50362,
        317,
        4171,
        2243,
        318,
        319,
        262,
        3084,
        13,
        50256
      ],
      "text": " A blue cop is on the table.",
      "eos": true,
      "seconds": 0.774603458121419,
      "rtf": 0.34829292181718474,
      "score": {
        "substitutions": 1,
        "deletions": 0,
        "insertions": 0,
        "reference_words": 7,
        "errors": 1,
        "wer": 0.14285714285714285
      }
    },
    {
      "id": "1-clean",
      "condition": "clean",
      "audio": "1-clean.wav",
      "reference": "The workshop opens at nine and closes at six.",
      "snr_db": null,
      "mix_gain": 1.0,
      "samples": 47104,
      "duration": 2.944,
      "audio_sha256": "6d23d3f10bb7c91ce9ba9470544ed974e649b0ca9281fe02c7ac7781d159099f",
      "feature_shape": [
        1,
        80,
        3000
      ],
      "token_ids": [
        50257,
        50362,
        383,
        20243,
        9808,
        379,
        860,
        290,
        20612,
        379,
        718,
        13,
        50256
      ],
      "text": " The workshop opens at 9 and closes at 6.",
      "eos": true,
      "seconds": 0.411659165751189,
      "rtf": 0.13982987967092017,
      "score": {
        "substitutions": 2,
        "deletions": 0,
        "insertions": 0,
        "reference_words": 9,
        "errors": 2,
        "wer": 0.2222222222222222
      }
    },
    {
      "id": "1-snr20",
      "condition": "snr20",
      "audio": "1-snr20.wav",
      "reference": "The workshop opens at nine and closes at six.",
      "snr_db": 20,
      "mix_gain": 1.0,
      "samples": 47104,
      "duration": 2.944,
      "audio_sha256": "63fe9a5d8e01da7650b71eec4ea6f42a39521a30cf1e10b742958e042a19b3c3",
      "feature_shape": [
        1,
        80,
        3000
      ],
      "token_ids": [
        50257,
        50362,
        383,
        20243,
        9808,
        379,
        860,
        290,
        20612,
        379,
        718,
        13,
        50256
      ],
      "text": " The workshop opens at 9 and closes at 6.",
      "eos": true,
      "seconds": 0.9168854579329491,
      "rtf": 0.3114420713087463,
      "score": {
        "substitutions": 2,
        "deletions": 0,
        "insertions": 0,
        "reference_words": 9,
        "errors": 2,
        "wer": 0.2222222222222222
      }
    },
    {
      "id": "1-snr0",
      "condition": "snr0",
      "audio": "1-snr0.wav",
      "reference": "The workshop opens at nine and closes at six.",
      "snr_db": 0,
      "mix_gain": 0.9457802719448573,
      "samples": 47104,
      "duration": 2.944,
      "audio_sha256": "18ceb13ebb78184043888458ad9c9f9022f0398f861a6a287761a18ce841a5a5",
      "feature_shape": [
        1,
        80,
        3000
      ],
      "token_ids": [
        50257,
        50362,
        383,
        670,
        4074,
        9808,
        379,
        6164,
        290,
        20612,
        379,
        10171,
        13,
        50256
      ],
      "text": " The work ship opens at mine and closes at stake.",
      "eos": true,
      "seconds": 0.45584845915436745,
      "rtf": 0.15483982987580416,
      "score": {
        "substitutions": 3,
        "deletions": 0,
        "insertions": 1,
        "reference_words": 9,
        "errors": 4,
        "wer": 0.4444444444444444
      }
    },
    {
      "id": "2-clean",
      "condition": "clean",
      "audio": "2-clean.wav",
      "reference": "Please pick up order number three seven two.",
      "snr_db": null,
      "mix_gain": 1.0,
      "samples": 46592,
      "duration": 2.912,
      "audio_sha256": "455cf25f92b603f0d31812a2f4b686c353adce2b4bc0cdf6fc88d1db62b2c331",
      "feature_shape": [
        1,
        80,
        3000
      ],
      "token_ids": [
        50257,
        50362,
        4222,
        2298,
        510,
        1502,
        1271,
        513,
        4761,
        13,
        50256
      ],
      "text": " Please pick up order number 372.",
      "eos": true,
      "seconds": 0.26073320768773556,
      "rtf": 0.08953750264001908,
      "score": {
        "substitutions": 1,
        "deletions": 2,
        "insertions": 0,
        "reference_words": 8,
        "errors": 3,
        "wer": 0.375
      }
    },
    {
      "id": "2-snr20",
      "condition": "snr20",
      "audio": "2-snr20.wav",
      "reference": "Please pick up order number three seven two.",
      "snr_db": 20,
      "mix_gain": 1.0,
      "samples": 46592,
      "duration": 2.912,
      "audio_sha256": "cfb7050917e36433bd6010b7c8132aa101964a143dd16dfb4e84f8bd5c9cdfc4",
      "feature_shape": [
        1,
        80,
        3000
      ],
      "token_ids": [
        50257,
        50362,
        4222,
        2298,
        510,
        1502,
        1271,
        513,
        4761,
        13,
        50256
      ],
      "text": " Please pick up order number 372.",
      "eos": true,
      "seconds": 0.8352768342010677,
      "rtf": 0.28683957218443257,
      "score": {
        "substitutions": 1,
        "deletions": 2,
        "insertions": 0,
        "reference_words": 8,
        "errors": 3,
        "wer": 0.375
      }
    },
    {
      "id": "2-snr0",
      "condition": "snr0",
      "audio": "2-snr0.wav",
      "reference": "Please pick up order number three seven two.",
      "snr_db": 0,
      "mix_gain": 0.8786484600388911,
      "samples": 46592,
      "duration": 2.912,
      "audio_sha256": "8dc1f1f5564d68fbbabf43deb02ad9016c869817a225acfb7af2ba28f94e7a18",
      "feature_shape": [
        1,
        80,
        3000
      ],
      "token_ids": [
        50257,
        50362,
        4222,
        2298,
        510,
        1502,
        1271,
        38107,
        13,
        50256
      ],
      "text": " Please pick up order number 272.",
      "eos": true,
      "seconds": 0.9308865000493824,
      "rtf": 0.3196725618301451,
      "score": {
        "substitutions": 1,
        "deletions": 2,
        "insertions": 0,
        "reference_words": 8,
        "errors": 3,
        "wer": 0.375
      }
    },
    {
      "id": "3-clean",
      "condition": "clean",
      "audio": "3-clean.wav",
      "reference": "There are two red circles and one blue square.",
      "snr_db": null,
      "mix_gain": 1.0,
      "samples": 56064,
      "duration": 3.504,
      "audio_sha256": "4ed5c23170fc804a93db38c6400670d7bda3f7f4d1bc1efd7c5e1403b3928357",
      "feature_shape": [
        1,
        80,
        3000
      ],
      "token_ids": [
        50257,
        50362,
        1318,
        389,
        734,
        2266,
        13332,
        290,
        530,
        4171,
        6616,
        13,
        50256
      ],
      "text": " There are two red circles and one blue square.",
      "eos": true,
      "seconds": 0.24269483285024762,
      "rtf": 0.06926222398694282,
      "score": {
        "substitutions": 0,
        "deletions": 0,
        "insertions": 0,
        "reference_words": 9,
        "errors": 0,
        "wer": 0.0
      }
    },
    {
      "id": "3-snr20",
      "condition": "snr20",
      "audio": "3-snr20.wav",
      "reference": "There are two red circles and one blue square.",
      "snr_db": 20,
      "mix_gain": 1.0,
      "samples": 56064,
      "duration": 3.504,
      "audio_sha256": "ab89fefc552d544fbe8000902a6f9d8849ace73e47bf31122548b1b36f584ee7",
      "feature_shape": [
        1,
        80,
        3000
      ],
      "token_ids": [
        50257,
        50362,
        1318,
        389,
        734,
        2266,
        13332,
        290,
        530,
        4171,
        6616,
        13,
        50256
      ],
      "text": " There are two red circles and one blue square.",
      "eos": true,
      "seconds": 0.19816054217517376,
      "rtf": 0.056552666145882925,
      "score": {
        "substitutions": 0,
        "deletions": 0,
        "insertions": 0,
        "reference_words": 9,
        "errors": 0,
        "wer": 0.0
      }
    },
    {
      "id": "3-snr0",
      "condition": "snr0",
      "audio": "3-snr0.wav",
      "reference": "There are two red circles and one blue square.",
      "snr_db": 0,
      "mix_gain": 0.8996655831313927,
      "samples": 56064,
      "duration": 3.504,
      "audio_sha256": "1a5341027ac5bd8e77e3f9495a7015929ceed8a130cd143731071af77960c1b9",
      "feature_shape": [
        1,
        80,
        3000
      ],
      "token_ids": [
        50257,
        50362,
        1318,
        389,
        734,
        2266,
        13332,
        290,
        530,
        4171,
        6616,
        13,
        50256
      ],
      "text": " There are two red circles and one blue square.",
      "eos": true,
      "seconds": 0.15009925002232194,
      "rtf": 0.04283654395614211,
      "score": {
        "substitutions": 0,
        "deletions": 0,
        "insertions": 0,
        "reference_words": 9,
        "errors": 0,
        "wer": 0.0
      }
    },
    {
      "id": "silence",
      "condition": "silence",
      "audio": "silence.wav",
      "reference": "",
      "snr_db": null,
      "mix_gain": 1.0,
      "samples": 64000,
      "duration": 4.0,
      "audio_sha256": "f751fd7094be088dc785125e4a1a0222a17c2bd20469aa31728ca5e515ed5809",
      "feature_shape": [
        1,
        80,
        3000
      ],
      "token_ids": [
        50257,
        50362,
        345,
        50256
      ],
      "text": " you",
      "eos": true,
      "seconds": 0.09338274970650673,
      "rtf": 0.023345687426626682,
      "score": {
        "substitutions": 0,
        "deletions": 0,
        "insertions": 1,
        "reference_words": 0,
        "errors": 1,
        "wer": null
      }
    }
  ],
  "environment": {
    "python": "3.12.12",
    "platform": "macOS-26.6.2-arm64-arm-64bit",
    "torch": "2.14.1",
    "numpy": "2.5.4",
    "transformers": "4.57.3",
    "device": "cpu",
    "dtype": "float32",
    "threads": 4
  },
  "config": {
    "sample_rate": 16000,
    "max_new_tokens": 64,
    "do_sample": false,
    "num_beams": 1,
    "return_timestamps": false,
    "return_dict_in_generate": true,
    "vad": false,
    "tts_seed_start": 2026,
    "noise_seed_start": 3030
  },
  "weights": {
    "tts": {
      "bytes": 145227512,
      "sha256": "69cf8b651c1493f1801dfd2311c298d694a38357bc9a1e41f410491ea6f0e1be"
    },
    "asr": {
      "bytes": 151060136,
      "sha256": "db59695928ded6043adaef491a53ef4e12da9611184d77c53baa691a60b958ad"
    }
  },
  "tts_config": {
    "vocab_size": 38,
    "hidden_size": 192,
    "num_hidden_layers": 6,
    "num_attention_heads": 2,
    "window_size": 4,
    "use_bias": true,
    "ffn_dim": 768,
    "layerdrop": 0.1,
    "ffn_kernel_size": 3,
    "flow_size": 192,
    "spectrogram_bins": 513,
    "hidden_act": "relu",
    "hidden_dropout": 0.1,
    "attention_dropout": 0.1,
    "activation_dropout": 0.1,
    "initializer_range": 0.02,
    "layer_norm_eps": 1e-05,
    "use_stochastic_duration_prediction": true,
    "num_speakers": 1,
    "speaker_embedding_size": 0,
    "upsample_initial_channel": 512,
    "upsample_rates": [
      8,
      8,
      2,
      2
    ],
    "upsample_kernel_sizes": [
      16,
      16,
      4,
      4
    ],
    "resblock_kernel_sizes": [
      3,
      7,
      11
    ],
    "resblock_dilation_sizes": [
      [
        1,
        3,
        5
      ],
      [
        1,
        3,
        5
      ],
      [
        1,
        3,
        5
      ]
    ],
    "leaky_relu_slope": 0.1,
    "depth_separable_channels": 2,
    "depth_separable_num_layers": 3,
    "duration_predictor_flow_bins": 10,
    "duration_predictor_tail_bound": 5.0,
    "duration_predictor_kernel_size": 3,
    "duration_predictor_dropout": 0.5,
    "duration_predictor_num_flows": 4,
    "duration_predictor_filter_channels": 256,
    "prior_encoder_num_flows": 4,
    "prior_encoder_num_wavenet_layers": 4,
    "posterior_encoder_num_wavenet_layers": 16,
    "wavenet_kernel_size": 5,
    "wavenet_dilation_rate": 1,
    "wavenet_dropout": 0.0,
    "speaking_rate": 1.0,
    "noise_scale": 0.667,
    "noise_scale_duration": 0.8,
    "sampling_rate": 16000,
    "return_dict": true,
    "output_hidden_states": false,
    "torchscript": false,
    "dtype": "float32",
    "pruned_heads": {},
    "tie_word_embeddings": true,
    "chunk_size_feed_forward": 0,
    "is_encoder_decoder": false,
    "is_decoder": false,
    "cross_attention_hidden_size": null,
    "add_cross_attention": false,
    "tie_encoder_decoder": false,
    "architectures": [
      "VitsModel"
    ],
    "finetuning_task": null,
    "id2label": {
      "0": "LABEL_0",
      "1": "LABEL_1"
    },
    "label2id": {
      "LABEL_0": 0,
      "LABEL_1": 1
    },
    "task_specific_params": null,
    "problem_type": null,
    "tokenizer_class": null,
    "prefix": null,
    "bos_token_id": null,
    "pad_token_id": null,
    "eos_token_id": null,
    "sep_token_id": null,
    "decoder_start_token_id": null,
    "max_length": 20,
    "min_length": 0,
    "do_sample": false,
    "early_stopping": false,
    "num_beams": 1,
    "temperature": 1.0,
    "top_k": 50,
    "top_p": 1.0,
    "typical_p": 1.0,
    "repetition_penalty": 1.0,
    "length_penalty": 1.0,
    "no_repeat_ngram_size": 0,
    "encoder_no_repeat_ngram_size": 0,
    "bad_words_ids": null,
    "num_return_sequences": 1,
    "output_scores": false,
    "return_dict_in_generate": false,
    "forced_bos_token_id": null,
    "forced_eos_token_id": null,
    "remove_invalid_values": false,
    "exponential_decay_length_penalty": null,
    "suppress_tokens": null,
    "begin_suppress_tokens": null,
    "num_beam_groups": 1,
    "diversity_penalty": 0.0,
    "_name_or_path": "/Users/chongyi/.cache/huggingface/hub/models--facebook--mms-tts-eng/snapshots/c71de0fe7204c83f1c10820a7d696d0b450048ba",
    "transformers_version": "4.57.3",
    "model_type": "vits",
    "tf_legacy_loss": false,
    "use_bfloat16": false,
    "output_attentions": false
  },
  "eos_token_id": 50256,
  "asr_generation_config": {
    "max_length": 448,
    "max_new_tokens": null,
    "min_length": 0,
    "min_new_tokens": null,
    "early_stopping": false,
    "max_time": null,
    "stop_strings": null,
    "do_sample": false,
    "num_beams": 1,
    "use_cache": true,
    "cache_implementation": null,
    "cache_config": null,
    "return_legacy_cache": null,
    "prefill_chunk_size": null,
    "temperature": 1.0,
    "top_k": 50,
    "top_p": 1.0,
    "min_p": null,
    "typical_p": 1.0,
    "epsilon_cutoff": 0.0,
    "eta_cutoff": 0.0,
    "repetition_penalty": 1.0,
    "encoder_repetition_penalty": 1.0,
    "length_penalty": 1.0,
    "no_repeat_ngram_size": 0,
    "bad_words_ids": null,
    "renormalize_logits": false,
    "forced_bos_token_id": null,
    "forced_eos_token_id": null,
    "remove_invalid_values": false,
    "exponential_decay_length_penalty": null,
    "suppress_tokens": [
      1,
      2,
      7,
      8,
      9,
      10,
      14,
      25,
      26,
      27,
      28,
      29,
      31,
      58,
      59,
      60,
      61,
      62,
      63,
      90,
      91,
      92,
      93,
      357,
      366,
      438,
      532,
      685,
      705,
      796,
      930,
      1058,
      1220,
      1267,
      1279,
      1303,
      1343,
      1377,
      1391,
      1635,
      1782,
      1875,
      2162,
      2361,
      2488,
      3467,
      4008,
      4211,
      4600,
      4808,
      5299,
      5855,
      6329,
      7203,
      9609,
      9959,
      10563,
      10786,
      11420,
      11709,
      11907,
      13163,
      13697,
      13700,
      14808,
      15306,
      16410,
      16791,
      17992,
      19203,
      19510,
      20724,
      22305,
      22935,
      27007,
      30109,
      30420,
      33409,
      34949,
      40283,
      40493,
      40549,
      47282,
      49146,
      50257,
      50357,
      50358,
      50359,
      50360,
      50361
    ],
    "begin_suppress_tokens": [
      220,
      50256
    ],
    "sequence_bias": null,
    "token_healing": false,
    "guidance_scale": null,
    "watermarking_config": null,
    "num_return_sequences": 1,
    "output_attentions": false,
    "output_hidden_states": false,
    "output_scores": false,
    "output_logits": null,
    "return_dict_in_generate": false,
    "pad_token_id": 50256,
    "bos_token_id": 50257,
    "eos_token_id": 50256,
    "encoder_no_repeat_ngram_size": 0,
    "decoder_start_token_id": 50257,
    "is_assistant": false,
    "num_assistant_tokens": 20,
    "num_assistant_tokens_schedule": "constant",
    "assistant_confidence_threshold": 0.4,
    "prompt_lookup_num_tokens": null,
    "max_matching_ngram_size": null,
    "assistant_early_exit": null,
    "assistant_lookbehind": 10,
    "target_lookbehind": 10,
    "disable_compile": false,
    "low_memory": null,
    "penalty_alpha": null,
    "dola_layers": null,
    "diversity_penalty": 0.0,
    "num_beam_groups": 1,
    "constraints": null,
    "force_words_ids": null,
    "_from_model_config": false,
    "transformers_version": "4.57.3",
    "alignment_heads": [
      [
        1,
        0
      ],
      [
        2,
        0
      ],
      [
        2,
        5
      ],
      [
        3,
        0
      ],
      [
        3,
        1
      ],
      [
        3,
        2
      ],
      [
        3,
        3
      ],
      [
        3,
        4
      ]
    ],
    "forced_decoder_ids": [
      [
        1,
        50362
      ]
    ],
    "is_multilingual": false,
    "max_initial_timestamp_index": 50,
    "no_timestamps_token_id": 50362,
    "prev_sot_token_id": 50360,
    "return_timestamps": false
  },
  "feature_config": {
    "feature_size": 80,
    "sampling_rate": 16000,
    "padding_value": 0.0,
    "padding_side": "right",
    "return_attention_mask": false,
    "_processor_class": "WhisperProcessor",
    "feature_extractor_type": "WhisperFeatureExtractor",
    "n_samples": 480000,
    "nb_max_frames": 3000,
    "n_fft": 400,
    "hop_length": 160,
    "chunk_length": 30,
    "dither": 0.0
  }
}