Sample Rate Converter

Sample rate converter

Find the exact relationship between two audio sample rates: the conversion ratio and the rational L/M resampling factors (upsample by L, downsample by M). You can also convert a sample count from the source rate to the target rate. It is aimed at audio engineers and developers setting up resampling pipelines.

Why the L/M factors matter

When you resample audio in software, the algorithm conceptually operates in two steps: first upsample by L (insert L−1 zero samples between each original sample), apply a low-pass filter at half the original Nyquist frequency, then downsample by M (keep every Mth sample). The integer ratio L/M tells you the minimum computational cost of the resampling operation. A ratio of 160/147 for 44.1 → 48 kHz means the filter must operate at 160× the output rate before decimation — significantly more expensive than, say, the trivial 1/2 downsample from 96 kHz to 48 kHz.

This is why the “awkward” 44.1 kHz to 48 kHz conversion is notorious: the 160/147 ratio has large factors, and a polyphase filter bank must compute 160 distinct phases. High-quality SRC libraries (libsamplerate, Secret Rabbit Code, SoX) handle this automatically, but understanding the factors helps you predict processing cost and choose the right quality/speed tradeoff.

How it works

The ratio is simply target ÷ source. To get the integer resampling factors, the tool divides both rates by their greatest common divisor (GCD):

L = target ÷ gcd(source, target) M = source ÷ gcd(source, target)

Resampling then means inserting samples to upsample by L, then discarding to downsample by M. A converted sample count is original × ratio.

Example

Converting 44,100 Hz to 48,000 Hz:

By contrast, converting 48,000 Hz to 96,000 Hz is trivially cheap: the ratio is exactly 2, GCD is 48,000, L/M is 2/1 — just insert one zero between every sample and filter. Knowing whether your conversion is a simple integer ratio or an awkward fraction helps you plan your audio pipeline.

Common conversion pairs

Everything is computed locally in your browser; no audio is processed or uploaded.