some conversion functions are not implemented in pa_converters.c

[PortAudio-admin]

Issue created by @RossBencina

Conversion functions in src/common/pa_converters.c copy audio data between PortAudio native buffers and host buffers.

https://www.assembla.com/code/portaudio/subversion/nodes/1930/portaudio/trunk/src/common/pa_converters.c

Some conversion functions are marked IMPLEMENT ME and do not pass any audio, some are marked REVIEW and may not currently work as expected.

All IMPLEMENT_ME functions are dithered variants of existing non-dither functions, so the non-dithered versions can be used as a starting point.

See also /test/patest_converters.c

[PortAudio-admin]

Comment by @RossBencina

removed assignment, i'm not taking responsibility for this, but i'll help if i can

[PortAudio-admin]

Comment by @RossBencina

I havn't volunteered to fix this, but I have been working on some code to test the converters if anyone is interested please email me or the list.

[PortAudio-admin]

Comment by @RossBencina

converter test code is in /test/patest_converters.c

[PortAudio-admin]

Comment by @RossBencina

This ticket will be greatly simplified once we decide on this one:

#112 -- require clipping for dithering sample conversion functions?

[PortAudio-admin]

Comment by @philburk

I implemented the Float32 to UInt8 conversions so that the loopback test would pass. The loopback test uses Float32 as its primary internal data type. Still more to go.

[PortAudio-admin]

Comment by @RossBencina

TRAC migration: propagating ticket status from TRAC

[dechamps]

Hello from 2023. Just FYI, this is still an issue for a small number of dithered converters. Here's the output of patest_converters on latest master (4bccdfb):

= Sine wave in, something out =


== flags = paNoFlag ==

=== Output contains non-zero data ===
Key: . - pass, X - fail
{{{
in|  out:      f32     i32     i24     i16      i8     ui8
f32             .       .       .       .       .       .
i32             .       .       X       .       .       X
i24             .       .       .       .       .       X
i16             .       .       .       .       X       X
 i8             .       .       .       .       .       .
ui8             .       .       .       .       .       .
}}}

=== Combined dynamic range (src->dest->float32) ===
Key: Noise amplitude in dBfs, X - fail (either above failed or dest->float32 failed)
{{{
in|  out:      f32     i32     i24     i16      i8     ui8
 f32          -inf  -180.6  -138.5   -79.5   -31.5   -31.5
 i32        -144.5  -144.5     X     -85.9   -42.1     X
 i24        -138.5  -138.5  -138.5   -86.2   -42.1     X
 i16         -84.8   -84.8   -84.8   -84.8     X       X
  i8         -36.6   -36.6   -36.6   -36.6   -36.6   -36.6
 ui8         -36.4   -36.4   -36.4   -36.4   -36.4   -36.4
}}}

== flags = paClipOff ==

=== Output contains non-zero data ===
Key: . - pass, X - fail
{{{
in|  out:      f32     i32     i24     i16      i8     ui8
f32             .       .       .       .       .       .
i32             .       .       X       .       .       X
i24             .       .       .       .       .       X
i16             .       .       .       .       X       X
 i8             .       .       .       .       .       .
ui8             .       .       .       .       .       .
}}}

=== Combined dynamic range (src->dest->float32) ===
Key: Noise amplitude in dBfs, X - fail (either above failed or dest->float32 failed)
{{{
in|  out:      f32     i32     i24     i16      i8     ui8
 f32          -inf  -180.6  -138.5   -80.0   -31.7   -31.1
 i32        -144.5  -144.5     X     -85.8   -42.1     X
 i24        -138.5  -138.5  -138.5   -85.8   -42.1     X
 i16         -84.8   -84.8   -84.8   -84.8     X       X
  i8         -36.6   -36.6   -36.6   -36.6   -36.6   -36.6
 ui8         -36.4   -36.4   -36.4   -36.4   -36.4   -36.4
}}}

== flags = paDitherOff ==

=== Output contains non-zero data ===
Key: . - pass, X - fail
{{{
in|  out:      f32     i32     i24     i16      i8     ui8
f32             .       .       .       .       .       .
i32             .       .       .       .       .       .
i24             .       .       .       .       .       .
i16             .       .       .       .       .       .
 i8             .       .       .       .       .       .
ui8             .       .       .       .       .       .
}}}

=== Combined dynamic range (src->dest->float32) ===
Key: Noise amplitude in dBfs, X - fail (either above failed or dest->float32 failed)
{{{
in|  out:      f32     i32     i24     i16      i8     ui8
 f32          -inf  -187.2  -138.7   -84.8   -36.6   -36.6
 i32        -144.5  -144.5  -138.5   -90.3   -42.1   -42.1
 i24        -138.5  -138.5  -138.5   -90.3   -42.1   -42.1
 i16         -84.8   -84.8   -84.8   -84.8   -42.1   -42.1
  i8         -36.6   -36.6   -36.6   -36.6   -36.6   -36.6
 ui8         -36.4   -36.4   -36.4   -36.4   -36.4   -36.4
}}}

== flags = paClipOff | paDitherOff ==

=== Output contains non-zero data ===
Key: . - pass, X - fail
{{{
in|  out:      f32     i32     i24     i16      i8     ui8
f32             .       .       .       .       .       .
i32             .       .       .       .       .       .
i24             .       .       .       .       .       .
i16             .       .       .       .       .       .
 i8             .       .       .       .       .       .
ui8             .       .       .       .       .       .
}}}

=== Combined dynamic range (src->dest->float32) ===
Key: Noise amplitude in dBfs, X - fail (either above failed or dest->float32 failed)
{{{
in|  out:      f32     i32     i24     i16      i8     ui8
 f32          -inf  -187.2  -138.5   -84.8   -36.6   -36.6
 i32        -144.5  -144.5  -138.5   -90.3   -42.1   -42.1
 i24        -138.5  -138.5  -138.5   -90.3   -42.1   -42.1
 i16         -84.8   -84.8   -84.8   -84.8   -42.1   -42.1
  i8         -36.6   -36.6   -36.6   -36.6   -36.6   -36.6
 ui8         -36.4   -36.4   -36.4   -36.4   -36.4   -36.4
}}}

The missing dithered i16 → i8 and i16 → ui8 converters are notably causing #168.

[philburk]

@dechamps - Thanks for the reminder. I added this to the V19.8 milestone.

[RossBencina]

#841 was work towards Int32_To_UInt8_Dither but was never finished (closed as abandoned). Could be useful as a reference.