Update resampler selection text to note that libsamplerate is particularly faster n 32-bit builds

This commit is contained in:
Chris Cannam
2023-06-26 16:42:42 +01:00
parent df596f472e
commit 1eddafd7a7

View File

@@ -273,8 +273,14 @@ library code, please be sure to review the terms for those libraries
in `src/ext/speex/COPYING` and `src/ext/kissfft/COPYING` as applicable.
If you are proposing to package Rubber Band for a Linux distribution,
please select either the built-in FFT or FFTW, and either the built-in
resampler or libsamplerate.
please select either the built-in FFT (the default) or FFTW
(`-Dfft=fftw`), and either the built-in resampler (the default) or
libsamplerate (`-Dresampler=libsamplerate`).
On some configurations (e.g. 32-bit platforms) libsamplerate is much
faster than the built-in resampler, and it makes a reasonable choice
for platform builds for which you don't know the requirements of any
specific application.
### FFT libraries supported
@@ -324,11 +330,14 @@ Built-in -Dresampler=builtin -DUSE_BQRESAMPLER Default.
Intended to give high quality
for time-varying pitch shifts
in real-time mode.
Not the fastest option.
Not the fastest option;
especially slow in 32-bit
builds.
libsamplerate -Dresampler=libsamplerate -DHAVE_LIBSAMPLERATE Good choice in most cases.
High quality and usually a bit
faster than the built-in option.
High quality; usually a bit
faster than the built-in option;
much faster in 32-bit builds.
BSD-ish licence.
libspeexdsp -Dresampler=libspeexdsp -DHAVE_LIBSPEEXDSP Very fast.