Wav2vec2 Large Xlsr Vietnamese
基于facebook/wav2vec2-large-xlsr-53模型微调的越南语自动语音识别模型
下载量 22
发布时间 : 3/2/2022
模型简介
该模型是针对越南语优化的自动语音识别(ASR)模型,基于XLSR Wav2Vec2架构,使用Common Voice、FOSD和VIVOS数据集进行微调。
模型特点
多数据集微调
使用Common Voice、FOSD和VIVOS三个越南语数据集进行训练,提高模型适应性
16kHz采样率支持
优化处理16kHz采样率的语音输入
无需语言模型
可直接使用,无需额外语言模型支持
模型能力
越南语语音识别
自动语音转文本
使用案例
语音转写
越南语语音转录
将越南语语音内容转换为文本
在Common Voice越南语测试集上WER为49.59%
语音助手
越南语语音命令识别
用于越南语语音助手或智能家居设备的语音命令识别
🚀 Wav2Vec2-Large-XLSR-53-越南语
本项目基于 facebook/wav2vec2-large-xlsr-53 模型,使用 Common Voice、FOSD 和 VIVOS 越南语数据集进行微调。使用该模型时,请确保语音输入的采样率为 16kHz。
📋 基本信息
属性 | 详情 |
---|---|
模型类型 | 基于 XLSR 的 Wav2Vec2 越南语语音识别模型 |
训练数据 | Common Voice、FOSD(https://data.mendeley.com/datasets/k9sxg2twv4/4)、VIVOS(https://ailab.hcmus.edu.vn/vivos) |
评估指标 | 词错误率(WER) |
标签 | 音频、自动语音识别、语音、XLSR 微调周 |
许可证 | Apache-2.0 |
🚀 快速开始
本模型可直接使用(无需语言模型),具体使用方法如下。
💻 使用示例
基础用法
import torch
import torchaudio
from datasets import load_dataset
from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor
ENCODER = {
"ia ": "iê ",
"ìa ": "iề ",
"ía ": "iế ",
"ỉa ": "iể ",
"ĩa ": "iễ ",
"ịa ": "iệ ",
"ya ": "yê ",
"ỳa ": "yề ",
"ýa ": "yế ",
"ỷa ": "yể ",
"ỹa ": "yễ ",
"ỵa ": "yệ ",
"ua ": "uô ",
"ùa ": "uồ ",
"úa ": "uố ",
"ủa ": "uổ ",
"ũa ": "uỗ ",
"ụa ": "uộ ",
"ưa ": "ươ ",
"ừa ": "ườ ",
"ứa ": "ướ ",
"ửa ": "ưở ",
"ữa ": "ưỡ ",
"ựa ": "ượ ",
"ke": "ce",
"kè": "cè",
"ké": "cé",
"kẻ": "cẻ",
"kẽ": "cẽ",
"kẹ": "cẹ",
"kê": "cê",
"kề": "cề",
"kế": "cế",
"kể": "cể",
"kễ": "cễ",
"kệ": "cệ",
"ki": "ci",
"kì": "cì",
"kí": "cí",
"kỉ": "cỉ",
"kĩ": "cĩ",
"kị": "cị",
"ky": "cy",
"kỳ": "cỳ",
"ký": "cý",
"kỷ": "cỷ",
"kỹ": "cỹ",
"kỵ": "cỵ",
"ghe": "ge",
"ghè": "gè",
"ghé": "gé",
"ghẻ": "gẻ",
"ghẽ": "gẽ",
"ghẹ": "gẹ",
"ghê": "gê",
"ghề": "gề",
"ghế": "gế",
"ghể": "gể",
"ghễ": "gễ",
"ghệ": "gệ",
"ngh": "\x80",
"uyê": "\x96",
"uyề": "\x97",
"uyế": "\x98",
"uyể": "\x99",
"uyễ": "\x9a",
"uyệ": "\x9b",
"ng": "\x81",
"ch": "\x82",
"gh": "\x83",
"nh": "\x84",
"gi": "\x85",
"ph": "\x86",
"kh": "\x87",
"th": "\x88",
"tr": "\x89",
"uy": "\x8a",
"uỳ": "\x8b",
"uý": "\x8c",
"uỷ": "\x8d",
"uỹ": "\x8e",
"uỵ": "\x8f",
"iê": "\x90",
"iề": "\x91",
"iế": "\x92",
"iể": "\x93",
"iễ": "\x94",
"iệ": "\x95",
"uô": "\x9c",
"uồ": "\x9d",
"uố": "\x9e",
"uổ": "\x9f",
"uỗ": "\xa0",
"uộ": "\xa1",
"ươ": "\xa2",
"ườ": "\xa3",
"ướ": "\xa4",
"ưở": "\xa5",
"ưỡ": "\xa6",
"ượ": "\xa7",
}
def decode_string(x):
for k, v in list(reversed(list(ENCODER.items()))):
x = x.replace(v, k)
return x
test_dataset = load_dataset("common_voice", "vi", split="test[:2%]")
processor = Wav2Vec2Processor.from_pretrained("Nhut/wav2vec2-large-xlsr-vietnamese")
model = Wav2Vec2ForCTC.from_pretrained("Nhut/wav2vec2-large-xlsr-vietnamese")
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:", [decode_string(x) for x in processor.batch_decode(predicted_ids)])
print("Reference:", test_dataset["sentence"][:2])
评估用法
本模型可在 Common Voice 越南语测试数据上进行评估,示例代码如下:
import torch
import torchaudio
from datasets import load_dataset, load_metric
from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor
import re
ENCODER = {
"ia ": "iê ",
"ìa ": "iề ",
"ía ": "iế ",
"ỉa ": "iể ",
"ĩa ": "iễ ",
"ịa ": "iệ ",
"ya ": "yê ",
"ỳa ": "yề ",
"ýa ": "yế ",
"ỷa ": "yể ",
"ỹa ": "yễ ",
"ỵa ": "yệ ",
"ua ": "uô ",
"ùa ": "uồ ",
"úa ": "uố ",
"ủa ": "uổ ",
"ũa ": "uỗ ",
"ụa ": "uộ ",
"ưa ": "ươ ",
"ừa ": "ườ ",
"ứa ": "ướ ",
"ửa ": "ưở ",
"ữa ": "ưỡ ",
"ựa ": "ượ ",
"ke": "ce",
"kè": "cè",
"ké": "cé",
"kẻ": "cẻ",
"kẽ": "cẽ",
"kẹ": "cẹ",
"kê": "cê",
"kề": "cề",
"kế": "cế",
"kể": "cể",
"kễ": "cễ",
"kệ": "cệ",
"ki": "ci",
"kì": "cì",
"kí": "cí",
"kỉ": "cỉ",
"kĩ": "cĩ",
"kị": "cị",
"ky": "cy",
"kỳ": "cỳ",
"ký": "cý",
"kỷ": "cỷ",
"kỹ": "cỹ",
"kỵ": "cỵ",
"ghe": "ge",
"ghè": "gè",
"ghé": "gé",
"ghẻ": "gẻ",
"ghẽ": "gẽ",
"ghẹ": "gẹ",
"ghê": "gê",
"ghề": "gề",
"ghế": "gế",
"ghể": "gể",
"ghễ": "gễ",
"ghệ": "gệ",
"ngh": "\x80",
"uyê": "\x96",
"uyề": "\x97",
"uyế": "\x98",
"uyể": "\x99",
"uyễ": "\x9a",
"uyệ": "\x9b",
"ng": "\x81",
"ch": "\x82",
"gh": "\x83",
"nh": "\x84",
"gi": "\x85",
"ph": "\x86",
"kh": "\x87",
"th": "\x88",
"tr": "\x89",
"uy": "\x8a",
"uỳ": "\x8b",
"uý": "\x8c",
"uỷ": "\x8d",
"uỹ": "\x8e",
"uỵ": "\x8f",
"iê": "\x90",
"iề": "\x91",
"iế": "\x92",
"iể": "\x93",
"iễ": "\x94",
"iệ": "\x95",
"uô": "\x9c",
"uồ": "\x9d",
"uố": "\x9e",
"uổ": "\x9f",
"uỗ": "\xa0",
"uộ": "\xa1",
"ươ": "\xa2",
"ườ": "\xa3",
"ướ": "\xa4",
"ưở": "\xa5",
"ưỡ": "\xa6",
"ượ": "\xa7",
}
def decode_string(x):
for k, v in list(reversed(list(ENCODER.items()))):
x = x.replace(v, k)
return x
test_dataset = load_dataset("common_voice", "vi", split="test")
wer = load_metric("wer")
processor = Wav2Vec2Processor.from_pretrained("Nhut/wav2vec2-large-xlsr-vietnamese")
model = Wav2Vec2ForCTC.from_pretrained("Nhut/wav2vec2-large-xlsr-vietnamese")
model.to("cuda")
chars_to_ignore_regex = '[\\\+\@\ǀ\,\?\.\!\-\;\:\"\“\%\‘\”\�]'
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):
batch["sentence"] = re.sub(chars_to_ignore_regex, '', batch["sentence"]).lower()
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)
# Preprocessing the datasets.
# We need to read the aduio files as arrays
def evaluate(batch):
inputs = processor(batch["speech"], sampling_rate=16_000, return_tensors="pt", padding=True)
with torch.no_grad():
logits = model(inputs.input_values.to("cuda"), attention_mask=inputs.attention_mask.to("cuda")).logits
pred_ids = torch.argmax(logits, dim=-1)
batch["pred_strings"] = processor.batch_decode(pred_ids)
# decode_string: We replace the encoded letter with the initial letters
batch["pred_strings"] = [decode_string(x) for x in batch["pred_strings"]]
return batch
result = test_dataset.map(evaluate, batched=True, batch_size=8)
print("WER: {:2f}".format(100 * wer.compute(predictions=result["pred_strings"], references=result["sentence"])))
测试结果:49.59 %
🔧 训练信息
训练过程使用了 Common Voice 的 train
、validation
数据集,以及 FOSD 和 VIVOS 数据集。训练脚本可在 此处 找到。
📄 许可证
本项目采用 Apache-2.0 许可证。
Voice Activity Detection
MIT
基于pyannote.audio 2.1版本的语音活动检测模型,用于识别音频中的语音活动时间段
语音识别
V
pyannote
7.7M
181
Wav2vec2 Large Xlsr 53 Portuguese
Apache-2.0
这是一个针对葡萄牙语语音识别任务微调的XLSR-53大模型,基于Common Voice 6.1数据集训练,支持葡萄牙语语音转文本。
语音识别 其他
W
jonatasgrosman
4.9M
32
Whisper Large V3
Apache-2.0
Whisper是由OpenAI提出的先进自动语音识别(ASR)和语音翻译模型,在超过500万小时的标注数据上训练,具有强大的跨数据集和跨领域泛化能力。
语音识别 支持多种语言
W
openai
4.6M
4,321
Whisper Large V3 Turbo
MIT
Whisper是由OpenAI开发的最先进的自动语音识别(ASR)和语音翻译模型,经过超过500万小时标记数据的训练,在零样本设置下展现出强大的泛化能力。
语音识别
Transformers 支持多种语言

W
openai
4.0M
2,317
Wav2vec2 Large Xlsr 53 Russian
Apache-2.0
基于facebook/wav2vec2-large-xlsr-53模型微调的俄语语音识别模型,支持16kHz采样率的语音输入
语音识别 其他
W
jonatasgrosman
3.9M
54
Wav2vec2 Large Xlsr 53 Chinese Zh Cn
Apache-2.0
基于facebook/wav2vec2-large-xlsr-53模型微调的中文语音识别模型,支持16kHz采样率的语音输入。
语音识别 中文
W
jonatasgrosman
3.8M
110
Wav2vec2 Large Xlsr 53 Dutch
Apache-2.0
基于facebook/wav2vec2-large-xlsr-53微调的荷兰语语音识别模型,在Common Voice和CSS10数据集上训练,支持16kHz音频输入。
语音识别 其他
W
jonatasgrosman
3.0M
12
Wav2vec2 Large Xlsr 53 Japanese
Apache-2.0
基于facebook/wav2vec2-large-xlsr-53模型微调的日语语音识别模型,支持16kHz采样率的语音输入
语音识别 日语
W
jonatasgrosman
2.9M
33
Mms 300m 1130 Forced Aligner
基于Hugging Face预训练模型的文本与音频强制对齐工具,支持多种语言,内存效率高
语音识别
Transformers 支持多种语言

M
MahmoudAshraf
2.5M
50
Wav2vec2 Large Xlsr 53 Arabic
Apache-2.0
基于facebook/wav2vec2-large-xlsr-53微调的阿拉伯语语音识别模型,在Common Voice和阿拉伯语语音语料库上训练
语音识别 阿拉伯语
W
jonatasgrosman
2.3M
37
精选推荐AI模型
Llama 3 Typhoon V1.5x 8b Instruct
专为泰语设计的80亿参数指令模型,性能媲美GPT-3.5-turbo,优化了应用场景、检索增强生成、受限生成和推理任务
大型语言模型
Transformers 支持多种语言

L
scb10x
3,269
16
Cadet Tiny
Openrail
Cadet-Tiny是一个基于SODA数据集训练的超小型对话模型,专为边缘设备推理设计,体积仅为Cosmo-3B模型的2%左右。
对话系统
Transformers 英语

C
ToddGoldfarb
2,691
6
Roberta Base Chinese Extractive Qa
基于RoBERTa架构的中文抽取式问答模型,适用于从给定文本中提取答案的任务。
问答系统 中文
R
uer
2,694
98