Implementation:Datajuicer Data juicer AudioFFmpegWrappedMapper

Knowledge Sources	Datajuicer_Data_juicer
Domains	Data_Processing, Mapping
Last Updated	2026-02-14 16:00 GMT

Overview

Concrete tool for applying FFmpeg audio filters to audio files in a dataset provided by Data-Juicer.

Description

AudioFFmpegWrappedMapper is a mapper operator that wraps FFmpeg audio filters for flexible audio processing. It uses the ffmpeg-python library to apply a specified FFmpeg filter with custom keyword arguments and global arguments to each audio file in a sample. Processed audio files are saved to a configurable output directory, and the sample's source file paths are updated accordingly. If no filter name is provided, the audio files remain unmodified. It extends the Mapper base class.

Usage

Import when you need to apply arbitrary FFmpeg audio filters to audio files without writing custom operator code.

Code Reference

Source Location

• Repository: Datajuicer_Data_juicer
• File: data_juicer/ops/mapper/audio_ffmpeg_wrapped_mapper.py

Signature

@OPERATORS.register_module("audio_ffmpeg_wrapped_mapper")
class AudioFFmpegWrappedMapper(Mapper):
    def __init__(self,
                 filter_name: Optional[str] = None,
                 filter_kwargs: Optional[Dict] = None,
                 global_args: Optional[List[str]] = None,
                 capture_stderr: bool = True,
                 overwrite_output: bool = True,
                 save_dir: str = None,
                 *args, **kwargs):

Import

from data_juicer.ops.mapper.audio_ffmpeg_wrapped_mapper import AudioFFmpegWrappedMapper

I/O Contract

Inputs

Name	Type	Required	Description
filter_name	Optional[str]	No	FFmpeg audio filter name to apply. Default: None (no-op)
filter_kwargs	Optional[Dict]	No	Keyword arguments passed to the FFmpeg filter. Default: None
global_args	Optional[List[str]]	No	List arguments passed to the FFmpeg command-line. Default: None
capture_stderr	bool	No	Whether to capture stderr output. Default: True
overwrite_output	bool	No	Whether to overwrite existing output files. Default: True
save_dir	str	No	Directory to store generated audio files. If not specified, outputs are saved alongside inputs. Can also be set via DJ_PRODUCED_DATA_DIR environment variable

Outputs

Name	Type	Description
samples	Dict	Transformed samples with processed audio file paths updated

Usage Examples

YAML Configuration

process:
  - audio_ffmpeg_wrapped_mapper:
      filter_name: afade
      filter_kwargs:
        type: in
        duration: 3

Related Pages

• Environment:Datajuicer_Data_juicer_Python_Runtime_Environment