Added a 4-gram language model based on a 40M token social media corpus.

README.md

@@ -0,0 +1,55 @@
+---
+language: sv
+datasets:
+- common_voice
+- NST Swedish ASR Database
+- P4
+metrics:
+- wer
+tags:
+- audio
+- automatic-speech-recognition
+- speech
+license: cc0-1.0
+model-index:
+- name: Wav2vec 2.0 large VoxRex Swedish 4-gram
+---
+
+# KBLab's wav2vec 2.0 large VoxRex Swedish (C) with 4-gram model
+Training of the acoustic model is the work of KBLab. See [VoxRex-C](https://huggingface.co/KBLab/wav2vec2-large-voxrex-swedish) for more details. This repo extends the acoustic model with a social media 4-gram language model for boosted performance.
+
+## Model description
+VoxRex-C is extended with a 4-gram language model estimated from a subset extracted from [The Swedish Culturomics Gigaword Corpus](https://spraakbanken.gu.se/resurser/gigaword) from Språkbanken. The subset contains 40M words from the social media genre between 2010 and 2015.
+
+## How to use
+Audio should be downsampled to 16kHz.
+
+```python
+import torch
+import torchaudio
+from datasets import load_dataset
+from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor
+test_dataset = load_dataset("common_voice", "sv-SE", split="test[:2%]").
+processor = Wav2Vec2Processor.from_pretrained("KBLab/wav2vec2-large-voxrex-swedish")
+model = Wav2Vec2ForCTC.from_pretrained("KBLab/wav2vec2-large-voxrex-swedish")
+resampler = torchaudio.transforms.Resample(48_000, 16_000)
+# Preprocessing the datasets.
+# We need to read the aduio files as arrays
+def speech_file_to_array_fn(batch):
+    speech_array, sampling_rate = torchaudio.load(batch["path"])
+    batch["speech"] = resampler(speech_array).squeeze().numpy()
+    return batch
+test_dataset = test_dataset.map(speech_file_to_array_fn)
+inputs = processor(test_dataset["speech"][:2], sampling_rate=16_000, return_tensors="pt", padding=True)
+with torch.no_grad():
+    logits = model(inputs.input_values, attention_mask=inputs.attention_mask).logits
+predicted_ids = torch.argmax(logits, dim=-1)
+print("Prediction:", processor.batch_decode(predicted_ids))
+print("Reference:", test_dataset["sentence"][:2])
+```
+
+## Training procedure
+Text data for the n-gram model is pre-processed by removing characters not part of the wav2vec 2.0 vocabulary and uppercasing all characters. After pre-processing and storing each text sample on a new line in a text file, a [KenLM](https://github.com/kpu/kenlm) model is estimated. See [this tutorial](https://huggingface.co/blog/wav2vec2-with-ngram) for more details.
+
+## Evaluation results
+The model was evaluated on the full Common Voice test set version 6.1. VoxRex-C achieved a WER of 9.03% without the language model and 6.47% with the language model.

alphabet.json

@@ -0,0 +1 @@
+{"labels": ["", "<s>", "</s>", "\u2047", " ", "T", "E", "A", "N", "R", "S", "I", "L", "D", "O", "M", "K", "G", "U", "V", "F", "H", "\u00c4", "\u00c5", "P", "\u00d6", "B", "J", "C", "Y", "X", "W", "Z", "\u00c9", "Q", "8", "2", "5", "9", "1", "6", "7", "3", "4", "0", "'"], "is_bpe": false}

config.json

@@ -0,0 +1,83 @@
+{
+  "activation_dropout": 0.05,
+  "apply_spec_augment": true,
+  "architectures": [
+    "Wav2Vec2ForCTC"
+  ],
+  "attention_dropout": 0.1,
+  "bos_token_id": 1,
+  "codevector_dim": 256,
+  "contrastive_logits_temperature": 0.1,
+  "conv_bias": true,
+  "conv_dim": [
+    512,
+    512,
+    512,
+    512,
+    512,
+    512,
+    512
+  ],
+  "conv_kernel": [
+    10,
+    3,
+    3,
+    3,
+    3,
+    2,
+    2
+  ],
+  "conv_stride": [
+    5,
+    2,
+    2,
+    2,
+    2,
+    2,
+    2
+  ],
+  "ctc_loss_reduction": "mean",
+  "ctc_zero_infinity": true,
+  "diversity_loss_weight": 0.1,
+  "do_stable_layer_norm": true,
+  "eos_token_id": 2,
+  "feat_extract_activation": "gelu",
+  "feat_extract_dropout": 0.0,
+  "feat_extract_norm": "layer",
+  "feat_proj_dropout": 0.05,
+  "feat_quantizer_dropout": 0.0,
+  "final_dropout": 0.0,
+  "gradient_checkpointing": true,
+  "hidden_act": "gelu",
+  "hidden_dropout": 0.05,
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 4096,
+  "layer_norm_eps": 1e-05,
+  "layerdrop": 0.05,
+  "mask_channel_length": 10,
+  "mask_channel_min_space": 1,
+  "mask_channel_other": 0.0,
+  "mask_channel_prob": 0.0,
+  "mask_channel_selection": "static",
+  "mask_feature_length": 10,
+  "mask_feature_prob": 0.0,
+  "mask_time_length": 10,
+  "mask_time_min_space": 1,
+  "mask_time_other": 0.0,
+  "mask_time_prob": 0.05,
+  "mask_time_selection": "static",
+  "model_type": "wav2vec2",
+  "num_attention_heads": 16,
+  "num_codevector_groups": 2,
+  "num_codevectors_per_group": 320,
+  "num_conv_pos_embedding_groups": 16,
+  "num_conv_pos_embeddings": 128,
+  "num_feat_extract_layers": 7,
+  "num_hidden_layers": 24,
+  "num_negatives": 100,
+  "pad_token_id": 0,
+  "proj_codevector_dim": 256,
+  "transformers_version": "4.8.2",
+  "vocab_size": 46
+}

language_model/4gram.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:60c6f63f684d7bdffd77c0d0c5419745af253620ead85b842d45f60d2e9c8434
+size 1151612569

language_model/attrs.json

@@ -0,0 +1 @@
+{"alpha": 0.5, "beta": 1.5, "unk_score_offset": -10.0, "score_boundary": true}

preprocessor_config.json

@@ -0,0 +1,10 @@
+{
+  "do_normalize": true,
+  "feature_extractor_type": "Wav2Vec2FeatureExtractor",
+  "feature_size": 1,
+  "padding_side": "right",
+  "padding_value": 0,
+  "processor_class": "Wav2Vec2ProcessorWithLM",
+  "return_attention_mask": true,
+  "sampling_rate": 16000
+}

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7138b3f9c5700388ddbd8b08a194290529778aba95327e7445a621cbe24ef508
+size 1262106353

special_tokens_map.json

@@ -0,0 +1 @@
+{"bos_token": "<s>", "eos_token": "</s>", "unk_token": "<unk>", "pad_token": "<pad>"}

tokenizer_config.json

@@ -0,0 +1 @@
+{"unk_token": "<unk>", "bos_token": "<s>", "eos_token": "</s>", "pad_token": "<pad>", "do_lower_case": true, "word_delimiter_token": "|", "special_tokens_map_file": "/home/viktor.enzell/.cache/huggingface/transformers/0b261be51f4e64178898783357df2bcf65f8812f0093cae011abf9c9e2219a9c.9d6cd81ef646692fb1c169a880161ea1cb95f49694f220aced9b704b457e51dd", "tokenizer_file": null, "name_or_path": "KBLab/wav2vec2-large-voxrex-swedish", "tokenizer_class": "Wav2Vec2CTCTokenizer", "processor_class": "Wav2Vec2ProcessorWithLM"}

vocab.json

@@ -0,0 +1 @@
+{"<pad>": 0, "<s>": 1, "</s>": 2, "<unk>": 3, "|": 4, "T": 5, "E": 6, "A": 7, "N": 8, "R": 9, "S": 10, "I": 11, "L": 12, "D": 13, "O": 14, "M": 15, "K": 16, "G": 17, "U": 18, "V": 19, "F": 20, "H": 21, "Ä": 22, "Å": 23, "P": 24, "Ö": 25, "B": 26, "J": 27, "C": 28, "Y": 29, "X": 30, "W": 31, "Z": 32, "É": 33, "Q": 34, "8": 35, "2": 36, "5": 37, "9": 38, "1": 39, "6": 40, "7": 41, "3": 42, "4": 43, "0": 44, "'": 45}