8
votes

If you have audio encoded at 44100Hz that means you have 44100 samples per second. Does this mean 44100 samples/sec for a channel, or for all channels?

For example if a song is stereo and encoded at 44100Hz, are there 44100 sample/sec for both channels (aka: 22050 samples per channel), or does every channel have 44100 samples (aka: every second has 88200 samples, 44100 for channel 1 and 44100 for channel 2).

2
The answers below are both correct. In multichannel audio (including stereo) one sample with multiple channels is sometimes referred to as a "sample frame" or simply a "frame" and some people prefer the term "frame rate" to avoid this confusion, but, in either case, frame-rate and sample-rate are the same.Bjorn Roche
@BjornRoche I think audio frame-rate is how fast packet-chunks are read on average, unrelated to sample-rate. Each chunk may have a variable number of packets. Each packet has 1 sample per channel. Since packet-chunks are an encoding product, raw pcm has no packet-chunks but consists only of packets.jiggunjer

2 Answers

7
votes

Yes, the sample rate is independent of the number of channels, so e.g. CD quality audio is stereo 16 bits 44.1 kHz which means that there are two channels each sampled at 44.1 kHz, and the raw data rate is therefore 44100 * 2 * 16 = 1411200 bits/sec = 176400 bytes/sec.

5
votes

Each channel is 'sampled' separately, so you will have as much values per 'sample frequency' as there are channels.

Data rate for the PCM (uncompressed) audio is

sample_freq * channels * bits_per_sample / 8

Most common are 16 bit samples, so you'll have 44100 * 2 * 2 for your stereo recording at 44100 sample rate.