Change Video Speed Online
Change any video speed from 0.25x to 4x. With automatic audio pitch correction.
.mp4, .mov, .avi, .mkv · up to 100 MB
What you can do
Adjust your video speed instantly
100% private
Your video is processed in your browser. Never uploaded to any server.
Pitch correction
Audio is adjusted with the WSOLA algorithm to maintain natural pitch at any speed.
0.25x to 4x
From extreme slow motion to time-lapse. Eight available speeds.
Instant processing
No queues or waiting. Video is processed directly in your browser.
How it works
Three steps, no hassle
Upload your video
Drag or select your video file. Compatible with MP4, MOV, WebM, and more formats.
Choose the speed
Select from 0.25x (slow motion) to 4x (time-lapse). Audio pitch is corrected automatically.
Download the result
Your video at the new speed is processed in your browser and ready to download in seconds.
FAQ
Got questions?
Speed change does not affect visual video quality: the frames are exactly the same, they are just played back at a different cadence (effective fps). At 2x speed, a 30fps video plays as if it were 60fps perceptual. There is no image degradation or re-encoding of video frames. However, audio must be re-encoded with FFmpeg's atempo filter to adjust speed while maintaining pitch, which involves a small audio re-encoding step. This is inherent to any speed adjustment tool, including Adobe Premiere Pro and DaVinci Resolve when using their own time-stretching algorithms.
Without pitch correction, speeding up audio also raises its pitch (the 'chipmunk effect') and slowing it down lowers it (the 'slow deep voice' effect). FFmpeg uses the atempo filter for time-stretching: it modifies audio duration without changing pitch using the WSOLA algorithm (Waveform Similarity Overlap-Add). This algorithm finds similar segments in the audio signal and overlaps them to lengthen or shorten the signal in a perceptually natural way. The atempo filter accepts values between 0.5 and 2.0, so for extreme speeds like 4x you need to chain two instances: atempo=2.0,atempo=2.0. For 0.25x, two filters are also used: atempo=0.5,atempo=0.5.
The tool offers the most practically useful speeds: 0.25x (extreme slow motion, ideal for analyzing sports or technical movements), 0.5x (moderate slow motion, natural for content review), 0.75x (slightly slow, useful for language learning or music), 1x (original speed), 1.25x (slightly faster, very popular for podcasts and online courses), 1.5x (noticeably faster without losing comprehension), 2x (double speed, standard for educational content), and 4x (time-lapse, useful for long processes like coding or cooking tutorials).
Yes. Processed videos are exported in a format compatible with all major platforms. Instagram Reels, TikTok, and YouTube Shorts accept MP4 video with H.264 codec, which is the standard output format. Note that each platform has its own duration limits: Instagram Reels accepts up to 90 seconds, TikTok up to 10 minutes, YouTube Shorts up to 60 seconds. If you speed up a 4-minute video to 4x, the resulting 1-minute video is perfect for YouTube Shorts.
In FFmpeg, changing video speed is done by modifying the PTS (Presentation TimeStamps) of each frame. The setpts filter controls when each frame should be displayed. For 2x playback speed, the command is -filter:v setpts=0.5*PTS: each frame's timestamp is divided by 2, making all frames appear twice as fast. For 0.5x (slow motion), it is setpts=2.0*PTS: timestamps are doubled, slowing playback. The formula is setpts=(1/speed)*PTS. This filter is different from frame interpolation (like FFmpeg's minterpolate or After Effects Optical Flow), which generates artificial intermediate frames to achieve ultra-smooth slow motion from video recorded at 30fps.
Real slow motion requires recording at high frame rates: a camera that records at 240fps and plays back at 30fps produces 8x slow motion with perfect fluidity and no interpolation. Software slow motion (like this tool does) takes a normally-recorded video and slows playback, simply showing each frame for longer. The result has fewer effective frames per second: a 30fps video at 0.25x shows 7.5 effective fps, which can appear as 'stuttery' movement compared to real slow motion. For professional slow motion content, high-speed cameras like the Sony RX100 VII (960fps), iPhone 15 Pro (4K at 120fps), or Phantom Flex4K (up to 1000fps) are the appropriate tools.
Change video speed: setpts, atempo, and time-stretching explained
Video speed adjustment involves two technically independent operations: modifying video cadence and adjusting audio duration. In FFmpeg, the setpts (Set Presentation TimeStamps) filter controls visual speed: it modifies the timestamps of each frame without altering its content. The formula setpts=PTS/speed accelerates the video (setpts=PTS/2 for 2x) while setpts=PTS*factor slows it down (setpts=PTS*2 for 0.5x). This process does not re-encode video frames, only modifies timing metadata, making it extremely computationally efficient. The complete processing chain in FFmpeg for a video at 2x is: ffmpeg -i input.mp4 -filter_complex '[0:v]setpts=0.5*PTS[v];[0:a]atempo=2.0[a]' -map '[v]' -map '[a]' output.mp4. This is exactly the logic executed by professional tools like Adobe Premiere Pro (with its Rate Stretch Tool), Final Cut Pro (with Retime), and DaVinci Resolve (with Speed Ramp), though with additional frame interpolation algorithms for smoother results in extreme slow motion.
For speeds beyond the 0.5x-2.0x range, FFmpeg's atempo filter requires chaining. The filter enforces the range limit [0.5, 2.0] by design, as the WSOLA algorithm produces audible artifacts outside this range. To achieve 4x speed, the correct chaining is atempo=2.0,atempo=2.0 (two 2x multiplications). For 0.25x, it is atempo=0.5,atempo=0.5. Intermediate values like 3x require: atempo=1.5,atempo=2.0 or atempo=3.0 if using the rubberband filter (more advanced). The WSOLA (Waveform Similarity Overlap-Add) algorithm was developed by Werner Verhelst and Marc Roelands in 1993. It works by identifying similar audio segments and overlapping them with crossfade to lengthen or shorten the signal: in acceleration mode, it removes redundant segments; in slow-down mode, it duplicates similar segments. The perceived naturalness of the result depends strongly on the type of audio: spoken voice tolerates time-stretching well up to 2x, tonal music starts sounding artificial above 1.5x, and percussive sound effects are the most sensitive to artifacts.
Practical use cases for speed adjustment are broad, ranging from entertainment to professional productivity. In online education, the 1.25x-1.5x speed is statistically the most popular: a 2019 MIT study on edX courses found that students who consume content at 1.5x retain information equally as well as at 1x, provided the content is clear. For content creators on TikTok and Instagram Reels, time-lapse at 4x-8x of creative processes (painting, cooking, construction) is one of the highest-engagement formats. In sports analysis, slow motion at 0.25x-0.5x allows analyzing movement technique, biomechanics, and error detection in jumps, throws, and golf swings. For programmers and streamers, speeding up long coding or gaming session recordings creates timelapse videos that compress hours into minutes. In the cinematographic realm, overcranking (recording at more fps than played back) for artistic slow motion uses the same mathematical principle but implemented in hardware: cameras like the Phantom Flex4K record at 1000fps for 24fps playback, creating 41x slow motion with perfect fluidity.