mumospee_librispeech / script /process_audio.py
jayliqinzhang's picture
Upload script/process_audio.py with huggingface_hub
111e2ad verified
from mutagen.flac import FLAC
from mutagen.mp3 import MP3
import io
import ast
def _get_audio_duration(bytes, file_type):
"""
Get the flac file duration.
"""
# Load the byte data into a BytesIO object
data = io.BytesIO(bytes)
if file_type == "flac":
# Load the bytes data
audio = FLAC(data)
if file_type == "mp3":
audio = MP3(data)
# Get the duration in seconds
duration = audio.info.length
return str(duration)
# ---
# Have to make two seperate functions to process audio string,
# in order to handle the dask partition properly.
def process_audio_string_path(audio_str):
audio_dict = ast.literal_eval(audio_str)
return audio_dict["path"]
def process_audio_string_duration(audio_str):
audio_dict = ast.literal_eval(audio_str)
path = audio_dict["path"]
a_type = path.split(".")[1]
try:
duration = _get_audio_duration(audio_dict["bytes"], a_type)
return duration
except Exception as e:
print(f"Get error {e}")
return "n/a"
def process_audio_partition(partition):
"""
Process the audio column from the dataframe to get the path and duration.
"""
# Perform operations on each partition as if it were a Pandas DataFrame
partition['path'] = partition['audio'].apply(process_audio_string_path)
partition['duration'] = partition['audio'].apply(process_audio_string_duration)
return partition
def process_audio_column(result_df):
# Extra steps to hanlde audio column.
meta = result_df.head(0) # Use the structure of the original DataFrame and add the new column
meta['path'] = 'string'
meta['duration'] = 'string'
result_df = result_df.map_partitions(process_audio_partition, meta=meta)
result_df = result_df.drop(columns=['audio'])
return result_df