pyflink.multimodal.expression.AudioExpressionAccessor

class AudioExpressionAccessor(expression)

Audio operations available on a DataFrame expression.

These methods use the expression on the left side as the audio input. For example, col("audio_bytes").audio.decode() is the expression-accessor form of audio_decode(col("audio_bytes")).

Examples::

>>> from pyflink.dataframe import col
>>> df = df.with_column("waveform", col("audio_bytes").audio.decode())
>>> df = df.with_column(
...     "speech_ready",
...     col("waveform").audio.standardize(sample_rate=16000, channels=1))

Methods

asr_whisper(*[, language, task, model, ...])	Run Whisper ASR over waveform values.
concat(*[, concurrency])	Concatenate an array of waveform rows.
convert_format(*[, format, on_error, ...])	Convert encoded audio to another encoded format.
decode(*[, on_error, max_decoded_bytes, ...])	Decode encoded audio into waveform rows.
detect_language(*[, top_k, model, ...])	Detect likely spoken language candidates for waveform values.
detect_speech(*[, aggressiveness, frame_ms, ...])	Detect speech activity ranges in waveform values.
duration(*[, concurrency])	Return audio duration in seconds for this expression.
duration_filter(*[, min_seconds, ...])	Check whether audio duration is within second bounds.
encode(*[, format, concurrency])	Encode waveform rows to audio bytes.
is_valid(*[, validation, concurrency])	Check whether this expression contains valid audio.
metadata(*[, concurrency])	Probe audio metadata for this expression.
resample(*, sample_rate[, concurrency])	Resample waveform values to a target sample rate.
silence_detection(*, threshold_db[, ...])	Detect low-amplitude silence ranges in waveform values.
size_filter(*[, min_bytes, max_bytes, ...])	Check whether encoded audio size is within byte bounds.
split_by_duration(*, segment_duration_ms[, ...])	Split audio by fixed duration.
split_by_speech(speech_activity, *[, ...])	Split audio using caller-provided speech activity ranges.
split_by_timestamp(timestamps, *[, ...])	Split audio by explicit timestamp ranges.
standardize(*[, sample_rate, channels, ...])	Standardize waveform sample rate and channel count.