Packed (interlaced) audio data

[lp35]

Hi,

Currently using Portaudio for retrieving audio signal, I saw that ltc decoder does not support interlaced data (for 16bits for instance). I had to create my own template.

I have made a working prototype for s16i template. Any chance to see this update upstream?

[x42]

Interleaved (not interlaced)?

With Portaudio the easiest is to just use the paNonInterleaved PaSampleFormat. Portaudio can also directly provide paUInt8.

I'm not opposed to adding adding more formats, but I don't really see the point to add all possible combinations. Support for float and 16bit was added "make the common case fast".

I have no interested to write code for other cases, it is easy to convert/de-interleave at application level and call ltc_decoder_write(), but you're very welcome to file a pull-request.

I suppose adding a stride parameter to the LTCWRITE_TEMPLATE define can work and simply skip over samples (and the application provides a buffer that is already offset to the given channel).

How did you implement it?

[lp35]

Yes I was meaning interlaced, paNonInterleaved in portaudio misled me. Thanks!

1. Concerning your portaudio suggestion, my pipeline need to use interlaced data, I can't add paNonInterleaved.

2. I agree with you regarding not supporting all the format the world can carry. However, depending on the rollout of Avoid buffer copy with ltc_decoder_write_ template #31, I think it could be a good thing to move the template LTCWRITE_TEMPLATE from ltc.c to ltc.h, so the users can implement their own conversion without recompiling ltc.

3. Agree on stride parameter.

4. Here you can see my implementation for short data interlaced:
   https://github.com/lp35/libltc/blob/master/src/ltc.c#L97
   Note: I really like how you did implement the template, on a coding style perspective.

[lp35]

Hi,

I re-up this question, I am ready to invest some time for this. Any thought on my last comments?

Thank youm BR

[x42]

LTCWRITE_TEMPLATE(s16i, short, 128 + (((short*)buf)[(copyStart + i)*2] >> 8))

would be for interleaved (not interlaced) stereo only. I suppose the * 2 should become a parameter alike:

ltc_decoder_write_interleaved_s16(LTCDecoder *d, short *buf, size_t size, size_t stride, ltc_off_t posinfo)

Then again, I'm probably going to regret this. The current 8,16, 32bit conversion functions are already overkill and converting sample-formats is not the purpose of this lib. The libraries copies data onto the stack for conversion and doing this in the lib cannot be more efficient than application code.