Change the sample rate of audio without aliasing
This audio sample rate converter resamples music, voice recordings and sound effects to the rate your project needs: 48 kHz for a video timeline, 44.1 kHz for a CD master or streaming upload, 16 kHz for speech recognition, 8 kHz for a phone system, or 88.2 and 96 kHz for a hi-res workflow. Drop in WAV, FLAC, MP3, M4A, Ogg or Opus files, pick a rate and download the result in the same format, tags intact.
Resampling is easy to do badly: dropping samples or joining neighbours with straight lines leaves harsh aliasing and dull highs. This tool uses a long windowed-sinc filter that keeps the band below about 90% of the new Nyquist frequency flat and pushes the rest more than 90 dB down.
Everything runs on your device — Mediabunny (MPL-2.0) decodes; libFLAC (BSD), LAME and libavcodec (LGPL) and libvorbis (BSD) encode — so unreleased mixes and client files never leave it.
How to use the Audio Sample Rate Converter
- Drop one or more audio files on the upload area. Each file lists its format, length, sample rate, bit depth, channels and bitrate.
- Choose the New sample rate, for example 48 kHz when the file is going into a video editor.
- Leave Output format on “Same as original”, or pick WAV, FLAC, MP3, AAC or Ogg Vorbis. For WAV and FLAC choose a Bit depth and keep TPDF dither ticked.
- Press Resample audio; Cancel stops at any time.
- Compare the Before and After players and the table of properties, then download the file, or all of them as a ZIP.
Audio Sample Rate Converter features
A proper anti-aliasing resampler
A polyphase windowed-sinc filter (Kaiser window, 64 zero crossings each side) works with the exact rate ratio, such as 160/147 for 44.1 → 48 kHz, so no timing error builds up over long files.
Every common rate
Nine standard rates from 8 to 96 kHz — telephone to studio — from any input rate the browser can decode.
Bit depth and dither
Save WAV as 16-bit, 24-bit or 32-bit float and FLAC as 16- or 24-bit, with optional triangular (TPDF) dither when the word length shrinks.
Keeps your format and tags
An MP3 stays an MP3, a FLAC stays a FLAC, and title, artist, album and cover art are copied across where the format supports them.
Batch and ZIP
Resample a folder of samples or stems in one go.
Before and after
Listen to both versions and compare their properties side by side.
Which sample rate should you use?
The sample rate is how many times per second the waveform is measured. The highest frequency a file can hold, the Nyquist frequency, is half its sample rate.
| Sample rate | Highest frequency kept | Typical use |
|---|---|---|
| 8 kHz | 4 kHz | Landline telephony, IVR prompts, intercoms |
| 16 kHz | 8 kHz | Speech recognition, VoIP “HD voice”, voice assistants |
| 22.05 kHz | 11.025 kHz | Small voice files, retro games, quick previews |
| 32 kHz | 16 kHz | Broadcast links, MiniDV and some radio systems |
| 44.1 kHz | 22.05 kHz | CD audio, music streaming and most MP3s |
| 48 kHz | 24 kHz | Video, film, DAWs, podcasts recorded on cameras |
| 88.2 / 96 kHz | 44.1 / 48 kHz | Hi-res releases and studio processing headroom |
How the resampler avoids aliasing
When you lower the sample rate, anything above the new Nyquist frequency must be removed first, or it folds back into the audible range as a false tone. A 15 kHz cymbal partial in a 44.1 kHz file that is crudely converted to 22.05 kHz reappears at 22.05 − 15 = 7.05 kHz. Here the filter cuts in at 95% of the lower Nyquist frequency and reaches roughly 100 dB of attenuation by the Nyquist frequency itself, so that alias is buried far below the 16-bit noise floor. The built-in self-test checks exactly this case.
Upsampling uses the same filter to remove the mirror images that appear between the old and new Nyquist frequencies. It cannot add detail: a 44.1 kHz recording converted to 96 kHz still has nothing above 22 kHz, so only convert up when a device, plug-in or delivery specification requires it.
Bit depth and dither
- 16-bit gives about 96 dB of dynamic range — the CD standard and plenty for finished music and speech.
- 24-bit gives about 144 dB and is the norm for recording and editing, where levels are still being changed.
- 32-bit float (WAV only) cannot clip during later processing and stores the resampler’s output without rounding.
- TPDF dither adds a tiny amount of noise before rounding so quiet fades turn into smooth hiss instead of gritty distortion. Leave it on unless the file will be measured sample by sample.
Format limits to know about
- MP3 supports 8 to 48 kHz, and its bitrate ceiling falls at lower rates: up to 320 kbps from 32 kHz, up to 160 kbps at 16–24 kHz and lower again at 8–12 kHz.
- Opus always runs at 48 kHz, so Opus files are saved as WAV when another rate is chosen.
- The FLAC encoder used here supports 8, 16, 22.05, 24, 32, 44.1, 48, 88.2 and 96 kHz and above; 11.025 kHz FLAC files are saved as WAV.
- Need fewer channels as well? Run the result through the Stereo to Mono Converter, or shrink a lossy file with the Audio Bitrate Converter.
Audio Sample Rate Converter FAQ
How do I convert 44.1 kHz to 48 kHz?
Drop the file here, choose 48 kHz as the new sample rate and press Resample audio. The tool uses the exact 160/147 ratio between the two rates, so the length stays sample-accurate and video stays in sync.
Does changing the sample rate reduce quality?
Lowering the rate removes everything above the new Nyquist frequency — 8 kHz audio sounds like a phone call — but this audio sample rate converter otherwise keeps the passband flat. Raising the rate keeps the sound the same and only makes the file larger.
Will the pitch or speed change?
No. Resampling recalculates the samples for the same duration, so pitch and tempo stay the same; pitch only shifts when audio is played at the wrong rate.
What sample rate is best for YouTube, podcasts and video?
48 kHz is the video and broadcast standard and is safe for YouTube and most podcast hosts. 44.1 kHz is the music standard. Speech-to-text services often prefer 16 kHz mono.
Should I use dither?
Yes when saving 16-bit, especially for quiet or fading material. It replaces rounding distortion with a low, even hiss about 96 dB down. Turn it off only for test signals that must match sample values exactly.