Resample supported audio to 8, 16, 22.05, 32, 44.1, 48, or 96 kHz locally. The tool measures the source rate, uses FFmpeg’s reviewed resampler path, preserves duration and pitch, explains any simultaneous lossy re-encoding, and verifies the output duration and actual sample rate.
Unavailable rates are disabled for the selected output codec.
| Input | Output | |
|---|---|---|
| File size | - | - |
| Container | - | - |
| Codec | - | - |
| Average bitrate | - | - |
| Sample rate | - | - |
| Channels | - | - |
| Duration | - | - |
| Metadata in output | - | - |
| Processing time | - | - |
Sample rate is the number of samples represented per second. Resampling calculates a new sequence at the target rate while keeping playback duration and pitch unchanged. It is different from changing playback speed.
The page runs the conversion through FFmpeg and checks the produced stream. If the chosen container requires a new lossy encode, that consequence is shown separately from resampling.
Sample rate describes time resolution and the theoretical frequency range. Bitrate describes encoded data per second for formats such as MP3 or AAC. They are related in encoder constraints but are not interchangeable settings.
Lowering sample rate does not guarantee a proportional file-size reduction in a fixed-bitrate lossy output. For PCM WAV, sample rate directly multiplies payload size.
44.1 kHz is common in music distribution and general audio. 48 kHz is common in video, broadcast, conferencing, and production workflows. Neither is universally superior; the receiving workflow should decide.
Avoid unnecessary conversions when the source already meets the destination requirement. Every extra lossy encode can add generation loss even when the sample-rate change itself is well filtered.
Lower target rates have a lower Nyquist frequency, equal to half the sample rate. Frequencies above the new limit must be filtered before reducing the sampling grid or they can fold into audible aliases.
The tool uses FFmpeg’s swresample path with a larger filter and controlled cutoff. It does not drop samples with a hand-written loop. The exact audible result still depends on source content and codec. Filter design includes a transition band near Nyquist, so the theoretical frequency limit is not a brick-wall statement about every frequency.
Changing 16 kHz audio to 48 kHz creates more sample positions, but it cannot recreate genuine high-frequency information that was never captured or was removed. Interpolation produces a valid stream for the new rate, not a higher-resolution original.
Upsampling can still be required by an editor, speech pipeline, or combined media project. The warning is about the claim, not the compatibility use case.
Mono 16 kHz audio is common in speech recognition and telephony-oriented pipelines because it covers the frequency range needed by many voice systems with less PCM data. It is not a universal podcast or archive standard in practice.
Check the service specification before converting. If music, ambience, or spatial detail matters, a wider bandwidth and stereo layout may be appropriate.
When possible, the tool selects an output compatible with the detected source. MP3, AAC, and Vorbis still require lossy re-encoding to write a new sample rate. WAV and FLAC avoid lossy coding but can create large in-memory outputs.
The result compares source and target rates, theoretical Nyquist frequency, estimated size, actual technical fields, and duration tolerance. A failed tolerance check is reported rather than hidden. Fixed-bitrate output may stay close to the same size after resampling, while PCM size changes roughly in proportion to rate when duration, channels, and bit depth remain fixed.
Not in this workflow. It computes a new stream at the target rate while preserving duration and pitch, then verifies duration tolerance.
It is half the sample rate and represents the theoretical upper frequency limit for a sampled signal.
No. Upsampling cannot create real high-frequency detail absent from the source.
MP3, AAC, and Vorbis must be encoded again to write a new stream, which adds a lossy generation.
Use the rate required by the destination. 44.1 kHz is common for music; 48 kHz is common for video and broadcast workflows.
The tool uses FFmpeg’s swresample implementation with a larger filter and controlled cutoff rather than simple sample dropping.