Bit Depth vs. Bit Rate: Why 16-Bit Is Not 16 kbps
Three numbers get confused constantly because they all contain the word bit:
- Bit depth — bits per sample, per channel. A unit of word length. Written as "16-bit," "24-bit."
- Sample rate — samples per second. A unit of frequency, written in Hz or kHz: 44,100 Hz, 48 kHz.
- Bit rate — bits per second of the finished stream. A unit of throughput, written in kbps or Mbps.
"16-bit" and "16 kbps" are therefore not variants of the same spec. One describes how finely each individual sample is quantized; the other describes how much data flows per second of playing time. A 16-bit file's bit rate depends on its sample rate and channel count and is three orders of magnitude larger than 16 kbps.
What bit depth controls
Bit depth sets how many discrete amplitude values each sample can take. The federal digitization guidelines define it as the word length used to encode each sample, and tie it directly to the dynamic range the encoding can represent — see the FADGI glossary entry for bit depth. Each additional bit roughly doubles the number of available amplitude steps, which corresponds to about 6 dB more range between the loudest encodable signal and the quantization noise floor.
Bit depth says nothing about frequency content. A 24-bit file is not "wider bandwidth" than a 16-bit one.
What sample rate controls
Sample rate sets how often amplitude is measured, and therefore the highest frequency the stream can represent — the Nyquist limit at half the sample rate. The FADGI definition of sampling rate states that basis explicitly. 44,100 samples per second gives a ceiling a little above 22 kHz, which is why that rate was chosen for a format intended to cover human hearing. We unpack that in what 44.1 kHz and 16-bit actually mean.
How the three connect
For uncompressed PCM — WAV, AIFF, the audio on a CD — bit rate is not an independent setting. It is the product of the other two numbers and the channel count:
`sample rate × bit depth × channels = bits per second`
44,100 × 16 × 2 = 1,411,200 bits per second, usually quoted as 1,411 kbps. The arithmetic and the file-size consequences are worked through in how to calculate audio bit rate and file size, and the Library of Congress recommended formats statement for audio works is a good illustration of why preservation specs are written as a rate-and-depth pair rather than a bit rate.
This is also why "convert 16-bit to kbps" has no answer by itself. 16-bit mono at 22.05 kHz, 16-bit stereo at 44.1 kHz, and 16-bit 5.1 at 48 kHz are all 16-bit and have wildly different bit rates.
Why compressed formats quote bit rate instead
Lossy codecs — MP3, AAC, Opus — discard information, so their output size is no longer a product of sample rate and word length. The encoder is given a target bit rate, and it decides internally how to spend those bits. That is why a lossy file's specification is "192 kbps AAC" and why asking for its bit depth is a category error: there is no fixed per-sample word length inside the compressed stream. A decoder reconstructs PCM samples at whatever depth the playback chain uses.
Lossless compression sits in between. A FLAC file encodes the exact original samples, so it still has a real bit depth and sample rate; what varies is the bit rate, because compression ratio depends on the content. Quiet or simple material compresses further than dense material, so the bit rate drifts across the file — the FLAC documentation describes the per-block encoding that produces that variability. See is FLAC really lossless for what "exact" means there.
Quick checks when reading a spec
- Units in kHz → sample rate. Units in kbps/Mbps → bit rate. A bare "16-bit" or "24-bit" → bit depth.
- If a format quotes a bit rate but no bit depth, it is almost certainly lossy.
- If a format quotes rate and depth but no bit rate, it is uncompressed and you can compute the bit rate yourself.
- A single bit rate for a lossless file is an average, not a constant.