Hi,

if I use FAAC to encode wav-files with sampling frequencies of 8000, 11025, 16000 and 22050 Hz, the encoder doubles the input sample rate, i.e. from 8000 input to 16000 output. Higher input rates result in correct ouput rates. Has anyone an idea?

I believe the decoder is doubling the sample rate in case SBR is used. I recall wondering the same thing a couple years ago. Some decoders didn't up-sample, others did. Don't ask me which did what, I don't remember.

Use winamp/foobar/mediainfo/etc. and view file details to see how it is really encoded.

Thanks for your reply. Did you mean ABR?

SBR = Spectral Band Replication. It is a technique in which the highest frequencies are not encoded, allowing for a reduced sample rate, but information is included to allow an approximation of the lost frequencies to be add at decode time.

i don't think FAAC supports SBR, so there must be something else.

An AAC decoder is allowed to double the samplerate even if no SBR is used when implicit SBR signalling is used, when the original samplerate is <= 24000. This is to make sure that for the decoder the actual samplerate it will output is exactly predictable without having to decode a frame first.