Flatten Audio Frames

@bleenco1 VersionTransform

Joins a [AudioFrame] batch head-to-tail into one AudioFrame, first unifying every frame to a common sample rate and channel layout so a mixed-format batch joins cleanly. Use when downstream needs one continuous segment but upstream emits a list of frames.

How it fits

Typical backends

Separate speakers into a per-speaker batch, flatten back into one frame, then transcribe.

Flatten a separated per-speaker batch into one continuous clip and write it to a file.

Caveats

I/O contractFrames in the input list may differ in sample rate and channel count. Each frame is remixed to the target channel count, then resampled to the target sample rate, before concatenation. Set target_sample_rate and target_channels to pin the output format; leave either unset to derive it from the batch as the highest sample rate and the maximum channel count, so nothing is lost by downsampling or downmixing.
I/O contractThe component concatenates by joining the underlying sample buffers; it does not cross-fade or smooth boundaries, so audible clicks or phase jumps between batches are preserved verbatim.
FallbackAn empty [AudioFrame] input returns a zero-sample AudioFrame. With a configured target_sample_rate / target_channels the empty frame carries that format; with neither configured there is no batch to derive a rate from, so it carries ``sample_rate = 0`` and one channel. No state is carried across calls.
CostMemory and per-call latency grow linearly with total audio length; resampling adds cost proportional to the rate change. Long sessions will OOM unless the upstream collector caps its buffer.

Versions

8d3909a1defaultlatest
Automated release
6/22/2026