Change Audio Sample Rate

A sample rate is how many times a second the audio was measured when it was recorded. 44.1 kHz means forty-four thousand one hundred measurements per second, and that number sets a hard ceiling on the highest pitch the file can hold — half the rate, so about 22 kHz, which is roughly the top of human hearing. Changing the rate rebuilds the whole waveform on a new grid of measurement points.

That rebuilding is where a converter is good or bad. Done well it is inaudible: the audible band comes through untouched and the artefacts introduced sit far below anything you could hear. Done badly it folds the frequencies it cannot carry back down into the music as a metallic ringing that no amount of later processing will remove. This page states its own numbers for both, and they are measured rather than claimed — see the quality section below.

Almost always because something downstream demands a specific number. Video is the common one: every broadcast and streaming delivery spec in use asks for 48 kHz, and a music file recorded at 44.1 kHz has to be converted before it can sit in the timeline. The reverse happens too — audio pulled out of a video at 48 kHz has to come back to 44.1 kHz to be released as music. Speech recognition and transcription services usually want 16 kHz mono and will do a worse job with anything else. And a mastering deliverable often specifies both a rate and a bit depth together.

• It does not re-encode a file that already matches If the rate, the format and the bit depth you asked for are what the file already has, there is nothing to convert — so the audio is copied straight through instead of being decoded and re-encoded. That is instant and perfectly lossless by construction. Measured on a ten-minute track it is 54 milliseconds against five and a half seconds, and the samples that come out are bit-for-bit the samples that went in. Every other online converter re-encodes it and charges you a generation of quality for nothing.
• It shows you what you have before you choose The file's real sample rate, bit depth, channel count and length are read and shown as soon as it lands, next to the target you are choosing. Most tools ask you to pick a number without telling you what you are picking it instead of.
• It never offers a setting it will quietly change MP3 cannot carry 96 kHz and FLAC cannot carry 32-bit float. Rather than accepting those and writing something else, the tool shows them crossed out with the reason — and the same applies to the settings your plan does not include, which are marked as such rather than silently downgraded.
• It reads the finished file back The rate and depth in the report are read out of the file that was written, not copied from the controls you set. That is the only way a report can tell you something you did not already know, and it is what catches the cases where a format quietly imposes its own limit.
• It goes to 192 kHz and 32-bit float Both families are covered end to end — 44.1, 88.2 and 176.4 kHz on one side, 48, 96 and 192 kHz on the other — and output runs to 32-bit float, which is the depth to use for anything that will be processed further because it needs no dither at all.
• It tells you which kind of conversion you asked for Halving 96 kHz to 48 kHz is an exact arithmetic operation. Going from 44.1 to 48 kHz is not — it is the awkward one every resampler is judged on. The tool names which of the two you are doing, because they are genuinely different jobs and no other web converter mentions it.

Sample rate conversion is normally sold on the name of the engine behind it. That tells you nothing, so here are the measurements instead. Each one is checked against a known-correct answer rather than against another program: a tone above the new file's ceiling must come out silent, and a tone inside the audible band must come out at exactly the level it went in at.

Unwanted artefacts, converting 96 kHz down to 44.1 kHz: 131 dB below the signal. For scale, a 24-bit file's own noise floor sits higher than that, so the conversion is quieter than the format carrying it. Flatness across the audible band, 100 Hz to 19 kHz: 0.00 dB — no measurable dip or lift anywhere. Distortion after a full round trip from 44.1 to 48 kHz and back: 135 dB down.

Speed, on the reference machine: ten minutes of stereo converted from 44.1 to 48 kHz in about five and a half seconds, and a ten-minute file that needs no conversion copied in 54 milliseconds.

Powered by optimised processing algorithms.

Bit depth is a separate thing from sample rate and controls how finely each individual measurement is recorded, which sets how quiet a sound the file can hold before it disappears into noise. 16-bit is the CD standard and is what most deliverables ask for. 24-bit is the working standard for recording and mastering. 32-bit float is not really about quietness at all — it can hold values beyond full scale without clipping, which makes it the right choice for anything that is going to be processed again.

Lowering the depth throws information away, and how you throw it away matters. Plain truncation produces a gritty distortion that follows the music and is most audible exactly where you least want it, on quiet fades and reverb tails. Dither replaces it with a steady, unrelated hiss that is far less objectionable even though there is technically more of it. Triangular dither is the safe, universally recommended default.

The two noise-shaped settings go further and move most of that noise up into the frequencies your hearing is least sensitive to. Measured on this tool, the shaped settings lift the noise above 12 kHz by about 24 to 26 dB relative to flat dither, which is exactly the trade they exist to make. Use them on a final 16-bit master; use plain triangular anywhere the file will be processed again. Raising the depth, or exporting to 32-bit float, needs no dither at all, and the control says so rather than sitting there pretending to do something.

• Music into a video edit 44.1 kHz to 48 kHz, WAV, 24-bit. The conversion every delivery spec asks for and the one this tool is fastest at telling you about.
• Audio out of a video, back to music Drop the MP4 straight in — the audio track is taken out and converted to 44.1 kHz without a separate extraction step.
• A 16-bit master from a 24-bit mix Keep the rate, drop the depth to 16, and use a noise-shaped dither. The tool confirms the finished file carries 16 bits rather than assuming it.
• Speech for transcription 16 kHz keeps everything a recogniser uses and makes the file a fraction of the size, which matters when a service charges by the megabyte.
• Standardising a folder of stems Queue them with Pro, set one rate and depth, convert in a single press, and take the whole set down as one archive.
• Checking what a file really is Drop it in and read the source line. The rate, depth, channel count and length are read from the file itself rather than from its name or extension.

• Convert the sample rate once, at the end, from the highest-quality source you have. Every conversion is a rebuild, and doing it twice is doing it twice.
• Set the bit depth in the same pass rather than in a separate step — dither belongs at the very end of the chain and applying it twice adds noise for nothing.
• Keep a lossless intermediate. Converting an MP3's rate means decoding and re-encoding it, which costs a generation; converting a WAV or FLAC costs nothing.
• Do not upsample to make a file sound better. It cannot add detail that was never recorded, and it multiplies the file size for no audible gain.
• When a downstream spec names a rate, match it exactly rather than going higher. Something later in the chain will convert it anyway, and probably worse.
• Check the report line on each finished file. It reads the rate and depth out of the file itself, which is where a format's own limits show up.

• Audio format converter
• Mono and stereo converter
• Audio normalizer
• LUFS loudness meter
• Volume booster
• All audio tools