Updates for hisstools library header-only

[AlexHarker]

This is the update for CLI to move to hisstools library header-only

[AlexHarker]

needed for flucoma/flucoma-core#265

[tremblap]

oh I should have meetings more often. things happen by magic!

[tremblap]

I'll listen to bufnmf null summing as audio out sanity test?

[tremblap]

I get a lot of these errors:

/research/dependencies/hisstools-src/include/audio_file/out_file.hpp:230:36: error: use of undeclared identifier 'endianness'
if (header_endianness() == endianness::little)
^

[tremblap]

(helping me to find the answer would be appreciated if you have the time to be verbose - I aim to improve my independence)

[tremblap]

ok i tried and it seems to be an error in the header file out_file.hpp that I cannot correct...

[tremblap]

should it be m_endianness instead?

[tremblap]

actually, I replaced endianness:: with endian_type:: and it works

[tremblap]

next bug:

FluidCLIWrapper.hpp:97:19: error: too few arguments to function call, single argument 'position' was not specified
file.seek();
~~~~~~~~~ ^
/Users/pa/Documents/documents@hudd/research/projects/fluid corpus navigation/research/dependencies/hisstools-src/include/audio_file/out_file.hpp:76:10: note: 'seek' declared here
void seek(uintptr_t position)

[tremblap]

ok so I missed the memo - it seems to work with the pull to the changes in HISStools and I'm chasing bugs that were sorted - but there is no note here that I should have pulled, hence the ghost chasing!

[tremblap]

ok real problem now: no more null-summing.

• download the CLI version, run HPSS with default settings, it nullsums in reaper
• with the latest compile of header only, it doesn't - I suspect bit resolution size is not behaving as it is all in the top end

[tremblap]

both are 32 bit float, so that is not the problem. Here is the output of afinfo on both versions - so it might be in the reading or the handling of the audio. I will stop pretending to troubleshoot but there is something not happy with file i/o

pa@tas ~ % afinfo /Users/pa/Desktop/h.wav
File: /Users/pa/Desktop/h.wav
File type ID: WAVE
Num Tracks: 1

Data format: 1 ch, 44100 Hz, Float32
no channel layout.
estimated duration: 10.293243 sec
audio bytes: 1815728
audio packets: 453932
bit rate: 1411200 bits per second
packet size upper bound: 4
maximum packet size: 4
audio data file offset: 44
optimized
source bit depth: F32

pa@tas ~ % afinfo /Users/pa/Desktop/h\ copy.wav
File: /Users/pa/Desktop/h copy.wav
File type ID: WAVE
Num Tracks: 1

Data format: 1 ch, 44100 Hz, Float32
no channel layout.
estimated duration: 10.293243 sec
audio bytes: 1815728
audio packets: 453932
bit rate: 1411200 bits per second
packet size upper bound: 4
maximum packet size: 4
audio data file offset: 44
optimized
source bit depth: F32

[tremblap]

I found the bug - there is a one sample offset in either input or output but I get 1 sample earlier with the new version

[tremblap]

[AlexHarker]

Thanks - is this a change of behaviour from the previous version?

[tremblap]

yes, the lined up file in the picture is yesterday's nightly. the only diffs are in core and here relating to the header-only hiss-magic

[AlexHarker]

Please redo all your cmake and test again - this should now be fixed (with core pulling in the fix)

[tremblap]

it all works thanks