pyflink.multimodal.expression.AudioExpressionAccessor.split_by_timestamp#

AudioExpressionAccessor.split_by_timestamp(timestamps: Union[pyflink.table.expression.Expression, List[Union[Dict[str, int], Sequence[int], pyflink.multimodal.expression._TimestampRow]], Tuple[Union[Dict[str, int], Sequence[int], pyflink.multimodal.expression._TimestampRow], ...]], *, segment_type: Literal['audio', 'ref'] = 'audio', concurrency: Optional[int] = None) → pyflink.dataframe.udf.DataFrameUDTFCall[source]#

Split audio by explicit timestamp ranges.

Equivalent to audio_split_by_timestamp(). See that function for full parameter details.

Parameters

timestamps – Timestamp ranges in milliseconds, either a Python literal sequence or a DataFrame column expression.
segment_type – "audio" for waveform segments or "ref" for lazy reference segments.
concurrency – Optional execution concurrency for this operation. None uses the framework default.

Returns

A DataFrame UDTF call for join_lateral. Alias the output as one segment column before consuming it downstream.

Examples::

>>> ranges = [{"start_ms": 100, "end_ms": 350}]
>>> segments = df.join_lateral(
...     col("waveform").audio.split_by_timestamp(
...         timestamps=ranges,
...     ).alias("segment")
... )
>>> segments = df.join_lateral(
...     col("waveform").audio.split_by_timestamp(
...         col("ranges"),
...     ).alias("segment")
... )