different_chunk_nums.py

import os
import pandas as pd
import librosa
import csv
import re

# Load the DataFrame from a CSV file
input_file = "deperson_13_files.csv"  # Replace with the actual file path
input_df = pd.read_csv(input_file)

output_dir = "./downloaded_audio"

chunks_to_remove = 5  # Number of chunks to remove

output = {
        'path': [],
        'segment': [],  # This will be used to track the segment number
        'start': [],
        'end': [],
}

def natural_sort_key(s):
    """
    Key function for natural sorting (sorts numbers in strings numerically).
    Example: "file2" < "file10" (unlike lexicographic sorting)
    """
    return [int(text) if text.isdigit() else text.lower()
            for text in re.split('([0-9]+)', s)]

def listdir_lsv(directory):
    """
    Returns files sorted like `ls -1v` (one per line, natural sort).
    """
    files = os.listdir(directory)
    return sorted(files, key=natural_sort_key)


def get_audio_duration(file_path):
    duration = librosa.get_duration(path=file_path)
    return duration

for index, row in input_df.iterrows():
    audio_path = row['path']
    current_segment_num = int(row['segment'])
    start_time = float(row['start'])

    basename = os.path.splitext(os.path.basename(audio_path))[0]
    segments_dir = os.path.join(output_dir, basename)
    segment_names = [f for f in listdir_lsv(segments_dir) if f.lower().endswith('.wav')]

    total_chunks_num = len(segment_names)

    if current_segment_num + chunks_to_remove >= total_chunks_num:
        end_time = get_audio_duration(os.path.join(output_dir, audio_path))
    else:
        final_segment_name = segment_names[current_segment_num + chunks_to_remove]
        final_segment_list = os.path.splitext(final_segment_name)[0].split("_")
        end_time = float(final_segment_list[3])

    output['path'].append(audio_path)
    output['segment'].append(current_segment_num)
    output['start'].append(start_time)
    output['end'].append(end_time)

    out_df = pd.DataFrame(output)
    out_df.to_csv(f'test_deperson_13_files_{chunks_to_remove}.csv', index=False,
                  quoting=csv.QUOTE_MINIMAL, sep=',')




















# if current_segment_num + chunks_to_remove >= total_chunks_num:
#     end_time = get_audio_duration(audio_file)
#     print(audio_file, start_time, end_time, sep=", ")
#     break
#
# final_segment_name = segment_names[current_segment_num + chunks_to_remove]
# final_segment_list = os.path.splitext(final_segment_name)[0].split("_")
# end_time = float(final_segment_list[3])
#
# audio = AudioSegment.from_file(audio_file)
#
# # Convert to milliseconds
# start_ms = int(start_time * 1000)
# end_ms = int(end_time * 1000)
#
# # Trim the audio
# trimmed = audio[:start_ms] + audio[end_ms:]
#
# print(audio_file, start_time, end_time, sep=", ")
# output['path'].append(base_name)
# output['segment'].append(current_segment_num)
# output['start'].append(start_time)
# output['end'].append(end_time)
# # Export trimmed audio
# # trimmed.export(os.path.join(out_dir, base_name + "_trimmed.wav"), format="wav")
# # print("Trimmed audio file to:", os.path.join(out_dir, base_name + "_trimmed.wav"))
# # sys.exit(0)