Recent projects need resampling algorithm, first find out a little, a few mainstream open source algorithm is, Speex / Opus / ffmpeg / sox
1. The first thing Speex, algorithm derived from the CCRMA (Center for Computer Research in Music and Acoustics) Music and Acoustics at Stanford University Center for Computer Research
A very independent algorithms, and supports the X86 ARM's NEON SIMD (SSE), also very simple, mainly on three functions init / process / destroy
Speex is easily extracted from the sourcecode inside.
2. When you see the article says Speex old men, has been replaced by Opus, Opus and there was ReSample resampling algorithm. And all owned by skype, called SILK
But unfortunately, Opus which resample only supports conversions between 8K / 12K / 16K / 24K / 48K and 8K / 12K / 16K, if it is to 48K from 44.1K MP3's
Opus recommend using the Speex algorithm.
3.ffmpeg there resampling, but he also said that if you are not satisfied, the proposed algorithm Sox's Resample
4.Sox
Just need to use is the 44100/22050 to 32000/16000, so you can not use the Opus.
The main research is Speex / ffmpeg / sox.
Speex start with the beginning, two free back again later study, no study details of the algorithm, saying only that application.