How to turn audio into a video

Give the recording a picture, keep the sound and make a file with both tracks.

Free toolAudio to VideoTurn audio into a video with cover art, a waveform or a spectrum. Choose a clip, add captions and save locally.Turn audio into video

Add a picture track, not just a different filename

Combine your audio with an image or animation, then export a video. Renaming an MP3 to MP4 does not do that. A container can hold sound without any picture.

YouTube’s supported-format guidance says audio-only MP3, WAV and PCM files cannot create a video upload. A still cover is enough to supply a picture track. Animation is optional.

Choose the excerpt and listen for a clean beginning and ending.
Pick a canvas shape, add artwork and keep the title readable at phone size.
Add a waveform or spectrum if it helps the recording. Import timed captions for speech.
Export, then play the saved file with sound before uploading it.

Choose a still cover, waveform or spectrum

Choose the visual for the recording’s purpose. Movement is not a requirement, and a visualiser does not improve audio quality.

Picture	What it communicates	Useful for
Still cover	Artwork and words stay in place. A progress line can show time.	A song, lesson or recording where the picture should stay quiet.
Waveform	Sample peaks show how the amplitude changes around the current moment.	Speech excerpts, voice notes and restrained podcast clips.
Spectrum	Frequency bands rise with energy at different pitches.	Music or a clip where the sound’s frequency changes are useful to see.
Ring spectrum	The same frequency idea arranged around the cover.	A compact composition centred on artwork.

Floi’s Waveform uses a moving 0.8-second window of sample peaks. Spectrum and Ring use a 2,048-sample Hann-windowed FFT with 48 logarithmic bands. These are visual representations, not calibrated loudness or frequency meters.

Keep captions aligned when you cut a clip

Keep the subtitle file timed to the original recording. Subtract the clip start only when creating a new clip-relative subtitle file. A renderer that reads source timestamps can do this during composition.

Consider a three-second excerpt from source time 2.00 to 5.00 seconds. The table shows the visible portion of three original cues. This worked example also covers a cue crossing either cut.

Cue in the original	Visible inside the excerpt	Time in the new video
1.50 to 2.50	2.00 to 2.50	0.00 to 0.50
3.00 to 4.00	3.00 to 4.00	1.00 to 2.00
4.50 to 5.50	4.50 to 5.00	2.50 to 3.00

In Floi, import the original SRT or VTT and set Clip start to 2 and Clip end to 5. Caption shift is a separate correction: +0.5 shows every cue half a second later. Do not shift by the clip start a second time.

Prepare the sound before making the picture

Listen to the excerpt on ordinary speakers. A moving waveform can look healthy while the recording remains too quiet or noisy.

Use an audio trimmer for a short excerpt from a long source.
Measure the recording with the loudness normaliser when matching clips from different sources.
Try the noise remover for distracting background noise. Check the voice after processing.
Create or correct captions before the final video export. Floi’s subtitle generator can make a timed file on your device.

Choose a browser studio or a desktop workflow

A browser studio suits one short recording and a simple design. A desktop editor or FFmpeg suits long sources, batch work and detailed control.

Method	Strength	Tradeoff
Floi	Local design preview, clip selection, artwork and imported captions without an account or watermark.	Five-minute sources, finite memory and native browser codec support. No automatic transcription in this studio.
Wayvl	Broader templates and on-device caption generation in its advertised workflow.	Its interface exposes more options. We tested a short WAV export, not transcription accuracy.
FilmScribe	Imported captions and substantially longer advertised audio support.	Our four-second trial reached a download link. We did not measure the claimed two-hour limit.
FFmpeg	Repeatable commands, batches and a wider range of codecs.	Installation and command-line work. You need to design the layout yourself.

Competitor observations date to 30 October 2026. Their limits and features can change. Local processing and watermark-free exports are available in other tools too.

Make a still-image video with FFmpeg

Loop the image, map the soundtrack and stop when the audio ends. This command fits the whole cover into a 1280 × 720 frame.

ffmpeg -loop 1 -i cover.png -i audio.wav \
  -vf "scale=1280:720:force_original_aspect_ratio=decrease,pad=1280:720:(ow-iw)/2:(oh-ih)/2,setsar=1" \
  -c:v libx264 -tune stillimage -pix_fmt yuv420p -r 30 \
  -c:a aac -b:a 192k -shortest -movflags +faststart output.mp4

Keep your source files. Encoding the audio as AAC is a lossy conversion. If compatible AAC already exists and the destination accepts it, an FFmpeg stream-copy workflow can avoid another audio encode.

Make a waveform video with FFmpeg

The showwaves filter draws a waveform from your audio. This example creates a square video and keeps the original soundtrack as AAC.

ffmpeg -i audio.wav \
  -filter_complex "[0:a]showwaves=s=720x720:mode=cline:colors=0xe9c780:rate=30,format=yuv420p[v]" \
  -map "[v]" -map 0:a -c:v libx264 -c:a aac -b:a 192k \
  -shortest -movflags +faststart waveform.mp4

The FFmpeg filter reference documents waveform modes. Use showfreqs when you want frequency bands instead. Neither command adds readable speech captions by itself.

What we checked in the saved files

We tested an original four-second, 48 kHz mono WAV on 30 October 2026. Its tone changes halfway through and contains four known impulses. FFprobe and decoded audio inspected the saved files on an Apple M1 Max MacBook Pro with 32 GB memory.

Method	Test	Observed file
Chrome 156 and automated WebKit	Square waveform, MP4	480 × 480, 120 picture frames, 4.000 seconds, H.264 and AAC. The impulses remained at 0.4, 1.4, 2.4 and 3.4 seconds.
Both browser engines	Portrait ring, WebM	480 × 600, 120 frames, 4.000 seconds, VP9 and Opus. The checked impulses retained their original positions.
Both browser engines	Source seconds 1 to 3, artwork and captions	854 × 480, 60 frames and 2.000 seconds. Source impulses at 1.4 and 2.4 became 0.4 and 1.4 seconds.
The FFmpeg still-cover command above	The same original WAV and a test image	1280 × 720, 120 picture frames and 4.000 seconds, with AAC sound.
The FFmpeg waveform command above	The same original WAV	720 × 720, 119 picture frames across 3.966667 seconds. The soundtrack and container lasted 4.000 seconds.

These are short synthetic checks, not measured device maxima. Automated WebKit is not a physical iPhone test. The FFmpeg waveform’s shorter picture track is another reason to check the end of a saved file.

Raw AAC and audio WebM can include codec or container padding. Our four-second fixtures opened as 4.032 and 4.008 seconds respectively. MP3 gapless metadata was respected in the tested file. A filename extension does not promise exact decoded duration.

Check the file you will actually share

Play the exported file from beginning to end with sound. Check the first word, last word and every caption near a cut.

Confirm that both a picture and a soundtrack are present.
Read the title and captions at the size people will see them.
Check artwork edges, pauses and the final frame.
Review the actual destination’s upload rules and preview. Its controls can cover text.

A successful local export does not prove a platform will accept it. Device memory, browser encoders and destination requirements can all differ.