Split CMSampleBufferRef containing Audio

Question

I'am splitting the recording into different files while recording...

The problem is, captureOutput video and audio sample buffers doesn't correspond 1:1 (which is logical)

- (void)captureOutput:(AVCaptureOutput *)captureOutput didOutputSampleBuffer:(CMSampleBufferRef)sampleBuffer fromConnection:(AVCaptureConnection *)connection

AUDIO START: 36796.833236847        | DURATION: 0.02321995464852608         | END: 36796.856456802
VIDEO START: 36796.842089239        | DURATION: nan         | END: nan
AUDIO START: 36796.856456805        | DURATION: 0.02321995464852608         | END: 36796.87967676
AUDIO START: 36796.879676764        | DURATION: 0.02321995464852608         | END: 36796.902896719
VIDEO START: 36796.875447239        | DURATION: nan         | END: nan
...

so i need to split the audio CMSampleBufferRef based on time and use the first segment for the first video and the second part of the buffer for the second video

It is possible to do things also with AVMutableComposition and AVAssetExportSession while exporting but the question is about the buffer level in captureOutput:, so the recorder file doesn't need more processing

---

Update:

Looks like 3 options, not successfully implemented yet

1) CMSampleBufferCopySampleBufferForRange

looks like CMSampleBufferCopySampleBufferForRange is the way to go, but i'am struggling to compute the last argument sampleRange...

2) CMSampleBufferCreateCopyWithNewTiming

quite lost using this one

3) looks like there is a way to trim the buffer by providing kCMSampleBufferAttachmentKey_TrimDurationAtStart, kCMSampleBufferAttachmentKey_TrimDurationAtEnd using the CMSetAttachment

Al

Were you able to fix that? what was the solution? – Merlon

nope, sorry, didnt figured this out :( – Al

did wen with a different approach a recorded the video and audio separatly – Al

CMSampleBufferCopySampleBufferForRange is the way to go, but only if you have interleaved (or mono) audio. If you have non interleaved, it's a CMBlockBufferCreateEmpty() for each piece, with CMBlockBufferAppendBufferReference() + CMAudioSampleBufferCreateWithPacketDescriptions() for each channel. – Thynne