Eating Sound Classification using Wav2Vec 2.0

How to use

Requirements

# requirement packages
!pip install git+https://github.com/huggingface/datasets.git
!pip install git+https://github.com/huggingface/transformers.git
!pip install torchaudio
!pip install librosa

Prediction

import torch
import torch.nn as nn
import torch.nn.functional as F
import torchaudio
from transformers import AutoConfig, Wav2Vec2FeatureExtractor

import librosa
import IPython.display as ipd
import numpy as np
import pandas as pd

device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
model_name_or_path = "m3hrdadfi/wav2vec2-base-100k-eating-sound-collection"
config = AutoConfig.from_pretrained(model_name_or_path)
feature_extractor = Wav2Vec2FeatureExtractor.from_pretrained(model_name_or_path)
sampling_rate = feature_extractor.sampling_rate
model = Wav2Vec2ForSpeechClassification.from_pretrained(model_name_or_path).to(device)

def speech_file_to_array_fn(path, sampling_rate):
    speech_array, _sampling_rate = torchaudio.load(path)
    resampler = torchaudio.transforms.Resample(_sampling_rate)
    speech = resampler(speech_array).squeeze().numpy()
    return speech


def predict(path, sampling_rate):
    speech = speech_file_to_array_fn(path, sampling_rate)
    inputs = feature_extractor(speech, sampling_rate=sampling_rate, return_tensors="pt", padding=True)
    inputs = {key: inputs[key].to(device) for key in inputs}

    with torch.no_grad():
        logits = model(**inputs).logits

    scores = F.softmax(logits, dim=1).detach().cpu().numpy()[0]
    outputs = [{"Label": config.id2label[i], "Score": f"{round(score * 100, 3):.1f}%"} for i, score in enumerate(scores)]
    return outputs

path = "clips_rd/gummies/gummies_6_04.wav"
outputs = predict(path, sampling_rate)

[
{'Label': 'aloe', 'Score': '0.0%'},
{'Label': 'burger', 'Score': '0.0%'},
{'Label': 'cabbage', 'Score': '0.0%'},
{'Label': 'candied_fruits', 'Score': '0.0%'},
{'Label': 'carrots', 'Score': '0.0%'},
{'Label': 'chips', 'Score': '0.0%'},
{'Label': 'chocolate', 'Score': '0.0%'},
{'Label': 'drinks', 'Score': '0.0%'},
{'Label': 'fries', 'Score': '0.0%'},
{'Label': 'grapes', 'Score': '0.0%'},
{'Label': 'gummies', 'Score': '99.8%'},
{'Label': 'ice-cream', 'Score': '0.0%'},
{'Label': 'jelly', 'Score': '0.1%'},
{'Label': 'noodles', 'Score': '0.0%'},
{'Label': 'pickles', 'Score': '0.0%'},
{'Label': 'pizza', 'Score': '0.0%'},
{'Label': 'ribs', 'Score': '0.0%'},
{'Label': 'salmon', 'Score': '0.0%'},
{'Label': 'soup', 'Score': '0.0%'},
{'Label': 'wings', 'Score': '0.0%'}
]

Evaluation

The following tables summarize the scores obtained by model overall and per each class.

Questions?

Post a Github issue from HERE.