linto-ai/sfeatpy

Library to extract MFCC features from audio signal

SFeatPy

Introduction

Python library to extract MFCC parameters.

Installation

pypi

pip install sfeatpy

From source

git clone https://github.com/linto-ai/sfeatpy.git
cd sfeatpy
./setup.py install

Usage

import sfeatpy
import numpy as np

rd_signal = np.random.random(16000)

res = sfeatpy.mfcc(rd_signal,           # audio signal
                   sample_rate,         # sample_rate -- Audio sampling rate (default 16000)  
                   window_length,       # window_length -- window size in sample (default 1024)  
                   window_stride,       # window_stride -- window stride in sample (default 512)  
                   fft_size,            # fft_size -- fft number of points (default 1024) 
                   min_freq,            # min_freq -- minimum frequency in hertz (default 20) 
                   max_freq,            # max_freq -- maximum frequency in hertz (default 7000) 
                   num_filter,          # num_filter -- number of MEL bins (default 40) 
                   num_coef,            # num_coef -- number of output coeficients (default 20) 
                   windowFun,           # windowFun -- window function: 0- None | 1- hamming (default 0) 
                   preEmp,              # preEmp -- preEmphasis factor ignored on None (default 0.97) 
                   keep_first_value     # keep_first_value -- if False discard first MFCC value (default False)
                   )
res.shape
> (30,20)

Limitations

• Values are not checked to keep the processing efficient.
• Works only on Mono-channel signal

Licence

This project is under aGPLv3 licence, feel free to use and modify the code under those terms. See LICENCE

Used libraries

• Numpy
• Scipy

sfeatpy's Issues

Seems very load heavy

import tensorflow.compat.v1 as tf
from tensorflow.python.ops import gen_audio_ops as audio_ops
import tensorflow as tf
import sounddevice as sd


rec_duration = 0.020
sample_rate = 16000
num_channels = 1
sd.default.never_drop_input= False
sd.default.latency= ('high', 'high')
sd.default.dtype= ('float32', 'float32')
sd.default.device = 'cap1'
upper_frequency_limit=7600
lower_frequency_limit=120
filterbank_channel_count=40
dct_coefficient_count=13

tf.compat.v1.disable_eager_execution()


def get_mfcc(waveform):
        # Run the spectrogram and MFCC ops to get a 2D audio: Short-time FFTs
        # background_clamp dims: [time, channels]
        spectrogram = audio_ops.audio_spectrogram(
            waveform,
            window_size=320,
            stride=160)
        # spectrogram: [channels/batch, frames, fft_feature]

        # extract mfcc features from spectrogram by audio_ops.mfcc:
        # 1 Input is spectrogram frames.
        # 2 Weighted spectrogram into bands using a triangular mel filterbank
        # 3 Logarithmic scaling
        # 4 Discrete cosine transform (DCT), return lowest dct_coefficient_count
        mfccs = audio_ops.mfcc(
            spectrogram=spectrogram,
            sample_rate=sample_rate,
            upper_frequency_limit=7600,
            lower_frequency_limit=120,
            filterbank_channel_count=40,
            dct_coefficient_count=13)
        # mfcc: [channels/batch, frames, dct_coefficient_count]
        # remove channel dim
        mfccs = tf.squeeze(mfccs, axis=0)
        return mfccs

def sd_callback(rec, frames, time, status):

    # Notify if errors
    if status:
        print('Error:', status)
    mfcc = get_mfcc(rec)
    print(mfcc)
    
# Start streaming from microphone
with sd.InputStream(channels=num_channels,
                    samplerate=sample_rate,
                    blocksize=int(sample_rate * rec_duration),
                    callback=sd_callback):
    while True:
        pass

Or

import sounddevice as sd
import numpy as np
import librosa


rec_duration = 0.020
sample_rate = 16000
num_channels = 1
sd.default.never_drop_input= False
sd.default.latency= ('high', 'high')
sd.default.dtype= ('float32', 'float32')
sd.default.device = 'cap1'


def get_mfcc(audio):
    audio = np.squeeze(audio, 1)    
    mfccs = librosa.feature.mfcc(y=audio, n_mfcc=13, sr=sample_rate, n_fft=320, hop_length=160, n_mels=40, fmin=60.0, fmax=76000.0, htk=False )
    return mfccs

def sd_callback(rec, frames, time, status):

    # Notify if errors
    if status:
        print('Error:', status)
    mfcc = get_mfcc(rec)
    print(mfcc)
    
# Start streaming from microphone
with sd.InputStream(channels=num_channels,
                    samplerate=sample_rate,
                    blocksize=int(sample_rate * rec_duration),
                    callback=sd_callback):
    while True:
        pass

sfeatpy in comparison seems heavy