Description #

Audio scoring operator - uses audiobox_aesthetics to score audio quality

Key features #

• Scores audio segments for quality using the audiobox_aesthetics model
• Provides four scoring dimensions: CE (coherence/listening engagement), CU (clarity/intelligibility), PC (production quality/compositional quality), PQ (perceived quality/subjective audio quality)
• Supports local and remote (tos://) audio files
• Applicable to audio quality assessment, audio filtering, quality control, and similar scenarios

Scoring dimension description #

• CE (coherence/listening engagement): Indicates whether the audio content is coherent and whether the audio is engaging from a subjective listening perspective
• CU (clarity/intelligibility): Indicates the clarity of the sound and the intelligibility of speech
• PC (production quality/compositional quality): Indicates the quality of recording, mixing, structure, and other production aspects
• PQ (perceived quality/subjective audio quality): Overall subjective listening quality

Operator parameters #

Input #

Output #

Structured result array, where each element contains the following fields:

• CE (float): Coherence/listening engagement score
• CU (float): Clarity/intelligibility score
• PC (float): Production quality/compositional quality score
• PQ (float): Perceived quality/subjective audio quality score

Audio processing failures return a structure containing a value of 0.0

Parameters #

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

Examples #

The following code demonstrates how to use daft to run the operator and score audio.

from __future__ import annotations

import os

import daft
from daft import col
from daft.las.functions.audio import AudioMetascore
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)

    samples = {"audio_path": [f"https://{TOS_TEST_DIR_URL}/public/archive/audio_metascore/sample.wav"]}
    df = daft.from_pydict(samples)
    df = df.with_column(
        "audio_metascore",
        las_udf(
            AudioMetascore,
            num_cpus=1,
            batch_size=1,
            concurrency=1,
        )(col("audio_path")),
    )

    df.show()
    # ╭────────────────────────────────┬────────────────────────────────────────────────────────────╮
    # │ audio_path                     ┆ audio_metascore                                            │
    # │ ---                            ┆ ---                                                        │
    # │ String                         ┆ Struct[CE: Float64, CU: Float64, PC: Float64, PQ: Float64] │
    # ╞════════════════════════════════╪════════════════════════════════════════════════════════════╡
    # │ https://las-public-data-qa.to… ┆ {CE: 5.909880638122559,                                    │
    # │                                ┆ CU: 6…                                                     │
    # ╰────────────────────────────────┴────────────────────────────────────────────────────────────╯