splitNile.py

#!/usr/bin/python
## Split audio files into chunks
## Daniel Pett 1/5/2020
__author__ = 'portableant'
## Tested on Python 2.7.16 - yes I know I need to upgrade.

import argparse
import os
import speech_recognition as sr
import csv
from csv import writer
from pydub import AudioSegment
from pydub.utils import which

AudioSegment.converter = which("ffmpeg")

from pydub.utils import make_chunks

r = sr.Recognizer()

parser = argparse.ArgumentParser(description='A script for splitting audio files into segments')

parser.add_argument('-m,', '--mp3', help='The MP3 convert folder', required=True)

parser.add_argument('-l', '--length', help='Length of chunk', required=True)
# An example would be --length 10000

# Parse arguments
args = parser.parse_args()

path = args.path
mp3 = args.mp3
destination = args.destination
length = args.length
chwav = '/Users/danielpett/Documents/githubProjects/fitz-transcribe-nile/chunk-wav/'
chmp3 = '/Users/danielpett/Documents/githubProjects/fitz-transcribe-nile/chunk-mp3/'

# Chunk mp3 files and convert to wav
for file in os.listdir(mp3):
    print('Now processing: ' + file)
    myaudio = AudioSegment.from_file(os.path.join(mp3,file), "mp3")
    # Make chunks of length specified
    chunk_length_ms = length
    chunks = make_chunks(myaudio, int(length))
    for i, chunk in enumerate(chunks):
        processedFileName = os.path.splitext(file)[0]
        chunk_name_mp3 = processedFileName + "_Chunk{0}.mp3".format(i)
        print("Now exporting: " , chunk_name_mp3)
        print(os.path.join(chmp3,chunk_name_mp3))
        chunk.export(os.path.join(chmp3,chunk_name_mp3), format="mp3")
        chunk_name_wav = processedFileName + "_Chunk{0}.wav".format(i)
        print("Now exporting: " , chunk_name_wav)
        chunk.export(os.path.join(chwav,chunk_name_wav), format="wav")

# Create the csv file for Pybossa import

transcriptions = []
for file in os.listdir(chwav):
    print('Now processing: ' + file)
    if file.endswith('.wav'):
        print("Source file " + file)
        sourceFile = os.path.join(chwav,file)
        audio = AudioSegment.from_file(sourceFile)
        duration =  audio.duration_seconds
        if(duration > 1):
            with sr.AudioFile(sourceFile) as source:
                # listen for the data (load audio to memory)
                audio_data = r.record(source)
                # recognize (convert from speech to text)
                text = r.recognize_google(audio_data)
                print('Output for ' + file + ': ' + text)
                extension = '.mp3'
                basefile = os.path.splitext(file)[0] + extension
                print(basefile)
                mp3file = os.path.join('https://fitz-audio-guide-micropasts.s3.eu-west-2.amazonaws.com/',basefile)
                transcriptions.append([mp3file,text])

print(transcriptions)

csvOutput = '/Users/danielpett/Documents/githubProjects/fitz-transcribe-nile/csv/transcriptions.csv'
file_exists = os.path.isfile(csvOutput)

with open (csvOutput, 'w') as csvfile:
    writer = csv.writer(csvfile , lineterminator='\n')
    writer.writerow(['track','currentTranscription'])
    for tup in transcriptions:
        writer.writerow(tup)

print('Good job my friend!')