Setting a Project Sample Rate and Reading It in Hertz
A session dialog offers 44.1, 48, 96 and 192. A converter datasheet talks about kSa/s. A codec register, an encoder argument and a resampler all want the same thing spelled out in full: samples per second. Every decision that follows the rate — where the highest reproducible tone lands, how much room the anti-alias filter has, what a plugin is really running at when it oversamples — is easier once the kilohertz label is expanded into hertz.
What the Rate Buys You Once It Is in Hertz
Nyquist is simply half the count
The filter needs somewhere to fall
Oversampling multiplies the figure
Working rate versus delivery rate
Working From a Kilohertz Spec to the Hertz Numbers
Whether the rate came from a session dialog, an interface control panel or the front page of a converter datasheet, the sequence is short.
Enter the rate exactly as the session shows it
Type 44.1, 48, 88.2, 96 or 192 into the kilohertz field and the samples-per-second figure appears beside it. Writing it as 44,1 works too, which matches how the rate is printed across much of Europe.
Halve it for the ceiling
The hertz figure is the one you halve. A 96 kHz project is 96 000 Hz, so nothing above 48 000 Hz can survive the sampling — useful when a synth or a modelled circuit is generating content far above hearing.
Paste the raw rate where a field wants hertz
The copy button returns 44100 with no unit and no spacing, which is what an encoder flag, a codec register or a resampler argument expects. Ctrl + C in the field does the same.
Reverse when a converter spec is already in hertz
Press the swap button (↔) to run Hz → kHz: an ADC advertised at 192 000 samples per second reads back as 192 kHz, the label the rest of the studio uses.
Standard Sample Rates and Their Nyquist Limits
Every rate you are likely to be offered, expanded into samples per second with the highest component it can carry and the place it is normally used.
| Rate | Samples per second | Nyquist limit | Where it is used |
|---|---|---|---|
| 8 kHz | 8 000 Hz | 4 000 Hz | Narrowband telephony and legacy voice codecs |
| 16 kHz | 16 000 Hz | 8 000 Hz | Wideband voice, speech recognition front ends |
| 32 kHz | 32 000 Hz | 16 000 Hz | Digital radio and some broadcast contribution links |
| 44.1 kHz | 44 100 Hz | 22 050 Hz | CD and most music distribution |
| 48 kHz | 48 000 Hz | 24 000 Hz | Video, broadcast and the usual session default |
| 88.2 kHz | 88 200 Hz | 44 100 Hz | Tracking destined for a 44.1 kHz master |
| 96 kHz | 96 000 Hz | 48 000 Hz | High-resolution music and film post |
| 176.4 kHz | 176 400 Hz | 88 200 Hz | Four times the 44.1 family, used in mastering chains |
| 192 kHz | 192 000 Hz | 96 000 Hz | Four times 48, archival captures and heavy modelling |
Notice the two families. Doubling stays inside a family — 44 100 to 88 200, 48 000 to 192 000 — while crossing between them is the awkward move that a resampler has to work at.
Useful When Planning a Session
Every standard rate in one pass
Type 44.1, then 48, then 96 — the samples-per-second side follows each one, so a rate table for a delivery spec can be filled in without clearing the field.
Back to a label the studio uses
The swap button runs Hz → kHz, which is the direction you want when a converter is specified at 192 000 samples per second and everyone else says 192.
Rates that paste into a command line
Copy hands back the digits alone, which is exactly the form an encoder argument, a script or a driver configuration file will accept.
Clocks far outside the audio band
The searchable dropdowns reach MHz, so a sigma-delta modulator clock running at 2.8 MHz can be compared with a 48 kHz output rate on the same page.
Sample Rate and Nyquist Questions
Where does the odd figure of 44 100 samples per second come from?
It came from storing digital audio on video tape. Early PCM adaptors wrote three samples per active video line, and the arithmetic works out the same on both television systems: 245 active lines × 60 fields × 3 = 44 100, and 294 × 50 × 3 = 44 100. That gave a comfortable margin above the 40 000 the Nyquist criterion demands for 20 000 Hz content, and the CD inherited it.
What is the Nyquist limit in hertz for the usual project rates?
Exactly half the samples-per-second figure: 22 050 Hz at 44.1 kHz, 24 000 Hz at 48 kHz, 48 000 Hz at 96 kHz and 96 000 Hz at 192 kHz. Only the first two ceilings sit anywhere near hearing; above them the extra room is for filter slopes and internal processing rather than for tones anybody perceives.
What does 4× oversampling in a plugin mean in hertz?
It means the plugin upsamples internally before doing its nonlinear work: inside a 48 000 Hz session, 4× puts it at 192 000 Hz. Clipping and saturation generate harmonics well above the session ceiling, and at the higher internal rate those products land below 96 000 Hz where the decimation filter can remove them instead of letting them fold back into the audible band.
Is a higher sample rate the same thing as more bits?
No — they answer different questions. The rate says how often a measurement is taken (44 100 or 96 000 times a second); the word length says how finely each measurement is graded. Each bit is worth about 6.02 dB of dynamic range, so 16 bits gives roughly 96 dB and 24 bits roughly 144 dB. Moving from 44 100 to 96 000 changes bandwidth and filter behaviour; it does nothing to the noise floor.
What actually happens when a 44.1 kHz file is moved to 48 kHz?
44 100 and 48 000 are in the ratio 147:160, so there is no whole-number relationship to exploit: the resampler has to interpolate new sample instants, and how cleanly it does that is down to the algorithm. Steps inside a family are far kinder — 88 200 to 44 100 and 192 000 to 48 000 are exact halvings. Where the delivery rate is known in advance, recording at it or at a multiple of it avoids the question entirely.
No comments yet. Be the first to comment!