Description #

Audio standardization module – standardizes audio to a specified format (sampling rate, channels, loudness, and more)

Key features #

• Supports resampling
• Supports channel unification (for example, converting to mono)
• Supports loudness normalization (target dBFS, with gain range limitation)
• Default input and output audio is in byte format

Usage scenarios #

• Unified processing of audio data from multiple sources
• Preprocessing for downstream ASR/TTS models
• Audio standardization step in multimodal content processing

Operator parameters #

Input #

Output #

Processed audio result; returns None if failed

Parameters #

If a parameter does not have a default value, it is required

Examples #

The following code demonstrates how to use the Daft operator AudioStandardization to standardize audio data to a specified format, including sampling rate, channels, and loudness.

# Copyright (c) Beijing Volcano Engine Technology Ltd.

from __future__ import annotations

import os

import daft
from daft import col
from daft.las.functions.audio.audio_standardization import AudioStandardization
from daft.las.functions.udf import las_udf

if __name__ == "__main__":
    TOS_TEST_DIR_URL = os.getenv("TOS_TEST_DIR_URL", "las-cn-beijing-public-online.tos-cn-beijing.volces.com")

    if os.getenv("DAFT_RUNNER", "native") == "ray":
        import logging

        import ray

        def configure_logging():
            logging.basicConfig(
                level=logging.INFO,
                format="%(asctime)s - %(name)s - %(levelname)s - %(message)s",
                datefmt="%Y-%m-%d %H:%M:%S",
            )
            logging.getLogger("tracing.span").setLevel(logging.WARNING)
            logging.getLogger("daft_io.stats").setLevel(logging.WARNING)
            logging.getLogger("DaftStatisticsManager").setLevel(logging.WARNING)
            logging.getLogger("DaftFlotillaScheduler").setLevel(logging.WARNING)
            logging.getLogger("DaftFlotillaDispatcher").setLevel(logging.WARNING)

        ray.init(dashboard_host="0.0.0.0", runtime_env={"worker_process_setup_hook": configure_logging})
        daft.set_runner_ray()

    daft.set_execution_config(actor_udf_ready_timeout=600)
    daft.set_execution_config(min_cpu_per_task=0)

    # Example input data
    samples = {"audio_path": [f"https://{TOS_TEST_DIR_URL}/public/archive/audio_standardization/.aac"]}

    # Construct Daft DataFrame
    df = daft.from_pydict(samples)

    # Apply AudioStandardization operator
    df = df.with_column(
        "standardized_audio",
        las_udf(
            AudioStandardization,
            construct_args={
                "target_sr": 16000,
                "target_channels": 1,
                "target_dbfs": -20.0,
                "target_gain_range": [-3.0, 3.0],
            },
            num_cpus=1,
            batch_size=1,
            concurrency=2,
        )(col("audio_path")),
    )

    df.show()
    # ╭────────────────────────────────┬────────────────────────────────╮
    # │ audio_path                     ┆ standardized_audio             │
    # │ ---                            ┆ ---                            │
    # │ String                         ┆ Binary                         │
    # ╞════════════════════════════════╪════════════════════════════════╡
    # │ https://las-public-data-qa.to… ┆ b"RIFF\xd0\x9a\x08\x00WAVEfmt… │
    # ╰────────────────────────────────┴────────────────────────────────╯