Sample rate explained
A digital recorder does not capture a continuous wave. It measures the air pressure at the mic thousands of times a second and stores each measurement as a number. The sample rate is how many measurements it takes per second.
The Nyquist limit in one sentence
To capture a frequency, you need at least two samples per cycle, so a sample rate can only represent frequencies up to half its value. That half is called the Nyquist frequency, after the Bell Labs engineer Harry Nyquist, and the theorem that proves the wave can then be reconstructed exactly is usually credited to Nyquist and Claude Shannon.
Watch aliasing happen
The grey curve is a real tone. The amber dots are the samples a recorder takes. Raise the tone past half the sample rate (the Nyquist limit) and the dots start tracing a different, lower tone: the amber dashed line. That phantom is aliasing.
Captured correctly: 3 < Nyquist limit 8
Real recorders put a low-pass filter in front of the converter so nothing above the Nyquist limit reaches it. At 44.1 kHz, that limit is 22.05 kHz, just above the roughly 20 kHz ceiling of young human hearing.
The same effect makes wagon wheels appear to spin backwards in old films: a camera shooting 24 frames per second is sampling the spokes too slowly. Photograph the Moon every 30 days and it barely seems to change phase, the sky's own aliasing; ahaboo narrates the full cycle those snapshots would miss.
Common sample rates
| Rate | Highest frequency | Where you meet it |
|---|---|---|
| 8 kHz | 4 kHz | Landline telephone |
| 16 kHz | 8 kHz | Speech recognition, VoIP "wideband", Bluetooth headset mics |
| 44.1 kHz | 22.05 kHz | Audio CD, most music streaming |
| 48 kHz | 24 kHz | Video, broadcast, most USB mics and browsers |
| 96 kHz | 48 kHz | High-resolution studio recording |
Why 44.1 kHz?
Human hearing tops out around 20 kHz (lower with age), so a rate a little over 40 kHz is needed, with some room for the anti-aliasing filter to roll off. The exact figure came from early digital recording, which stored audio on video tape: 44,100 fitted neatly into both the NTSC and PAL video line structures. Sony and Philips kept it for the CD in 1980.
Sample rate and your recordings
- Browsers usually record at 48 kHz. You can see the rate of any file or take in the audio editor's info line.
- For voice, recording above 48 kHz makes files larger without an audible gain.
- Speech-to-text models work at 16 kHz, which is why Audio to Text downsamples before sending: it cuts the upload by two-thirds with no loss in accuracy.
- File size grows in direct proportion to sample rate for WAV; see the calculator on the converter page.
Questions
What sample rate should I record at?
48 kHz is the standard for video and most podcast workflows; 44.1 kHz is the CD and music-streaming standard. Either is fine for voice. Higher rates such as 96 kHz mainly help if you plan heavy pitch or time manipulation.
Is 44.1 kHz or 48 kHz better?
Neither is audibly better. Pick the one that matches your destination (48 kHz for video, 44.1 kHz for music releases) to avoid a resampling step.
What is the difference between sample rate and bit depth?
Sample rate is how often the level is measured (sets the highest frequency). Bit depth is how finely each measurement is stored (sets the noise floor and dynamic range: about 6 dB per bit, so 16-bit ≈ 96 dB and 24-bit ≈ 144 dB).
Does VocalForge change my sample rate?
No. Recordings and files keep their original rate when edited and exported as WAV or MP3. OGG (Opus) export always uses 48 kHz because the Opus codec works at that rate internally.