Calculate sampling rate in Hz or kHz from sample count and duration, the rate needed for a signal’s highest frequency (Nyquist), or recording duration.

Use the per-channel frame count and total block duration.

Fractional counts describe timing equivalents, not complete acquired frames.

N / rate is block duration; first-to-last timestamp spacing is different.

PCM size settings

Default: 24 packed bits. Use container width, not only valid audio bits.

Default: 2; blank also uses 2. Supports 1–128 whole channels.

Sampling Rate Formula

The calculator uses one of three formulas depending on the solve mode you select.

Find Rate (samples and duration known):

fs = N / t

Twice-frequency boundary (highest signal frequency known):

fsboundary = 2 × fmax

Find Duration (samples and rate known):

t = N / fs
  • fs = sampling rate in samples per second (Hz)
  • N = frame count / samples per channel, not all interleaved channel values
  • t = uniform-rate block duration in seconds (N / fs), rather than the first-to-last sample timestamp difference
  • fmax = highest frequency component in the signal (Hz)
  • fsboundary = twice the highest frequency; a boundary rather than a general alias-free guarantee

For general real-valued baseband reconstruction, keep signal content below half the sampling rate and use a suitable anti-alias filter. Exactly 2 × fmax can lose amplitude or phase information. The calculator applies a chosen 100–1,000% of that boundary; its default 110% is an example assumption, not a universal safe margin. Filter requirements and hardware determine an appropriate rate. The duration formula assumes constant rate and uniform spacing.

Common Sampling Rates and Nyquist Limits

Illustrative reference values for real-valued baseband sampling. The last column is the mathematical fs/2 boundary, not a guaranteed captured bandwidth. These examples are not equipment, clinical ECG, or acquisition specifications; complex I/Q and bandpass sampling need their own analysis.

Application Typical Rate Rate/2 Boundary
Telephone voice8 kHz4 kHz
Wideband speech (VoIP)16 kHz8 kHz
CD audio44.1 kHz22.05 kHz
Studio / video audio48 kHz24 kHz
High-res audio96 / 192 kHz48 / 96 kHz
Vibration / ECG1 to 5 kHz500 Hz to 2.5 kHz
Ultrasonic sensors200 kHz to 1 MHz100 to 500 kHz
RF / SDR baseband2 to 20 Msps1 to 10 MHz
Highest Signal Frequency 2 × Frequency Boundary Example Target (2.2×)
100 Hz200 Hz220 Hz
1 kHz2 kHz2.2 kHz
10 kHz20 kHz22 kHz
20 kHz40 kHz44 kHz
100 kHz200 kHz220 kHz
1 MHz2 MHz2.2 MHz

Worked Example and FAQ

Example: You record a sensor and end up with 480,000 samples over 10 seconds. The rate is 480,000 ÷ 10 = 48,000 samples/s, or 48 kHz. Its fs/2 boundary is 24 kHz; general reconstruction requires content below that boundary and appropriate filtering.

What does 44.1 kHz allow for a 20 kHz signal band? The fs/2 boundary is 22.05 kHz, above the 20 kHz band edge. This leaves a transition region for a suitable filter; it does not establish a universal filter margin or explain the historical choice of the CD rate.

What happens if you sample below the Nyquist rate? Out-of-band components can fold into the captured spectrum as aliases. An anti-alias filter before the ADC and an appropriate rate prevent this loss; already aliased data generally cannot be uniquely repaired without additional information or reacquisition.

Is higher always better? No. At fixed stored bit depth, channels and duration, doubling the rate doubles raw PCM payload and bitrate. Actual processing cost and compressed file size depend on the system. Choose the rate and filtering for the signal and acquisition requirements.

How do you find samples if you know rate and duration? Multiply: N = fs × t. A 10-second block at 44.1 kHz contains 441,000 samples per channel. A fractional count is a timing equivalent rather than a complete acquired frame count. Packed PCM bytes use N × channels × stored bits per sample ÷ 8; headers, padding and compression are excluded. Use 32 stored bits for 24 valid bits held in a 32-bit container.