File size: 7,287 Bytes
634774b 06a825a 634774b 06a825a 634774b 06a825a 3ca9037 06a825a |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 |
---
language: is
datasets:
- language-and-voice-lab/samromur_milljon
tags:
- audio
- automatic-speech-recognition
- icelandic
- xlrs-53-icelandic
- iceland
- reykjavik
- samromur
license: cc-by-4.0
model-index:
- name: wav2vec2-large-xlsr-53-icelandic-ep30-967h
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: Samrómur (Test)
type: language-and-voice-lab/samromur_asr
split: test
args:
language: is
metrics:
- name: WER
type: wer
value: 7.698
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: Samrómur (Dev)
type: language-and-voice-lab/samromur_asr
split: validation
args:
language: is
metrics:
- name: WER
type: wer
value: 6.786
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: Samrómur Children (Test)
type: language-and-voice-lab/samromur_children
split: test
args:
language: is
metrics:
- name: WER
type: wer
value: 6.467
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: Samrómur Children (Dev)
type: language-and-voice-lab/samromur_children
split: validation
args:
language: is
metrics:
- name: WER
type: wer
value: 4.234
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: Malrómur (Test)
type: language-and-voice-lab/malromur_asr
split: test
args:
language: is
metrics:
- name: WER
type: wer
value: 6.631
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: Malrómur (Dev)
type: language-and-voice-lab/malromur_asr
split: validation
args:
language: is
metrics:
- name: WER
type: wer
value: 5.836
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: Althingi (Test)
type: language-and-voice-lab/althingi_asr
split: test
args:
language: is
metrics:
- name: WER
type: wer
value: 17.904
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: Althingi (Dev)
type: language-and-voice-lab/althingi_asr
split: validation
args:
language: is
metrics:
- name: WER
type: wer
value: 17.931
---
# wav2vec2-large-xlsr-53-icelandic-ep30-967h
The "wav2vec2-large-xlsr-53-icelandic-ep30-967h" is an acoustic model suitable for Automatic Speech Recognition in Icelandic. It is the result of fine-tuning the model [facebook/wav2vec2-large-xlsr-53](https://huggingface.co/facebook/wav2vec2-large-xlsr-53) for 30 epochs with 967 hours of Icelandic data collected by the [Language and Voice Laboratory](https://huggingface.co/language-and-voice-lab) through the platform [Samrómur](https://samromur.is/).
The specific data that was used to fine-tune the model is the corpus [Samrómur Milljón](https://huggingface.co/datasets/language-and-voice-lab/samromur_milljon), which is the result of the automatic verification of 1 million of recordings comming from the corpus ["Samromur Unverified 22.07"](http://hdl.handle.net/20.500.12537/265). It has to be pointed out that this model was trained with different data than our previous model [
wav2vec2-large-xlsr-53-icelandic-ep10-1000h ](https://huggingface.co/carlosdanielhernandezmena/wav2vec2-large-xlsr-53-icelandic-ep10-1000h).
The fine-tuning process was performed during July (2023) in the servers of the Language and Voice Laboratory (https://lvl.ru.is/) at Reykjavík University (Iceland) by [Carlos Daniel Hernández Mena](https://huggingface.co/carlosdanielhernandezmena).
# Evaluation
```python
import torch
from transformers import Wav2Vec2Processor
from transformers import Wav2Vec2ForCTC
#Load the processor and model.
MODEL_NAME="language-and-voice-lab/wav2vec2-large-xlsr-53-icelandic-ep30-967h"
processor = Wav2Vec2Processor.from_pretrained(MODEL_NAME)
model = Wav2Vec2ForCTC.from_pretrained(MODEL_NAME)
#Load the dataset
from datasets import load_dataset, load_metric, Audio
ds=load_dataset("language-and-voice-lab/samromur_children", split="test")
#Downsample to 16kHz
ds = ds.cast_column("audio", Audio(sampling_rate=16_000))
#Process the dataset
def prepare_dataset(batch):
audio = batch["audio"]
#Batched output is "un-batched" to ensure mapping is correct
batch["input_values"] = processor(audio["array"], sampling_rate=audio["sampling_rate"]).input_values[0]
with processor.as_target_processor():
batch["labels"] = processor(batch["normalized_text"]).input_ids
return batch
ds = ds.map(prepare_dataset, remove_columns=ds.column_names,num_proc=1)
#Define the evaluation metric
import numpy as np
wer_metric = load_metric("wer")
def compute_metrics(pred):
pred_logits = pred.predictions
pred_ids = np.argmax(pred_logits, axis=-1)
pred.label_ids[pred.label_ids == -100] = processor.tokenizer.pad_token_id
pred_str = processor.batch_decode(pred_ids)
#We do not want to group tokens when computing the metrics
label_str = processor.batch_decode(pred.label_ids, group_tokens=False)
wer = wer_metric.compute(predictions=pred_str, references=label_str)
return {"wer": wer}
#Do the evaluation (with batch_size=1)
model = model.to(torch.device("cuda"))
def map_to_result(batch):
with torch.no_grad():
input_values = torch.tensor(batch["input_values"], device="cuda").unsqueeze(0)
logits = model(input_values).logits
pred_ids = torch.argmax(logits, dim=-1)
batch["pred_str"] = processor.batch_decode(pred_ids)[0]
batch["sentence"] = processor.decode(batch["labels"], group_tokens=False)
return batch
results = ds.map(map_to_result,remove_columns=ds.column_names)
#Compute the overall WER now.
print("Test WER: {:.3f}".format(wer_metric.compute(predictions=results["pred_str"], references=results["sentence"])))
```
**Test Result**: 0.076
# BibTeX entry and citation info
*When publishing results based on these models please refer to:*
```bibtex
@misc{mena2023xlrs53icelandic30ep967h,
title={Acoustic Model in Icelandic: wav2vec2-large-xlsr-53-icelandic-ep30-967h.},
author={Hernandez Mena, Carlos Daniel},
url={https://huggingface.co/language-and-voice-lab/wav2vec2-large-xlsr-53-icelandic-ep30-967h},
year={2023}
}
```
# Acknowledgements
Thanks to Jón Guðnason, head of the Language and Voice Lab for providing computational power to make this model possible.
We also want to thank to the "Language Technology Programme for Icelandic 2019-2023" which is managed and coordinated by Almannarómur, and it is funded by the Icelandic Ministry of Education, Science and Culture. This model is an unexpected result of all the resources gathered by the Programme.
Special thanks to Björn Ingi Stefánsson for setting up the configuration of the server where this model was trained.
|