wav2vec2 large xlsr 53 gender recognition librispeech EU AI Act Compliance Profile

wav2vec2-large-xlsr-53-gender-recognition-librispeech

This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on Librispeech-clean-100 for gender recognition. It achieves the following results on the evaluation set:

• Loss: 0.0061
• F1: 0.9993

Compute your inferences

import os
import random
from glob import glob
from typing import List, Optional, Union, Dict

import tqdm
import torch
import torchaudio
import numpy as np
import pandas as pd
from torch import nn
from torch.utils.data import DataLoader
from torch.nn import functional as F
from transformers import (
    AutoFeatureExtractor,
    AutoModelForAudioClassification,
    Wav2Vec2Processor
)

class CustomDataset(torch.utils.data.Dataset):
    def __init__(
        self,
        dataset: List,
        basedir: Optional[str] = None,
        sampling_rate: int = 16000,
        max_audio_len: int = 5,
    ):
        self.dataset = dataset
        self.basedir = basedir

        self.sampling_rate = sampling_rate
        self.max_audio_len = max_audio_len

    def __len__(self):
        """
        Return the length of the dataset
        """
        return len(self.dataset)

    def __getitem__(self, index):
        if self.basedir is None:
            filepath = self.dataset[index]
        else:
            filepath = os.path.join(self.basedir, self.dataset[index])

        speech_array, sr = torchaudio.load(filepath)

        if speech_array.shape[0] > 1:
            speech_array = torch.mean(speech_array, dim=0, keepdim=True)

        if sr != self.sampling_rate:
            transform = torchaudio.transforms.Resample(sr, self.sampling_rate)
            speech_array = transform(speech_array)
            sr = self.sampling_rate

        len_audi

Your obligations as a wav2vec2 large xlsr 53 gender recognition librispeech deployer

Track your progress

Sign up to track obligations, receive deadline reminders, and generate audit packages.

Tags

Library

Base Model