A 30-second podcast audiogram costs about 2.2 cents from two ReelSDK calls: POST /v1/ffmpeg draws the waveform over the cover art at $0.02 per compute minute, and POST /v1/captions adds the words at $0.03 per output minute.
Make podcast audiograms and waveform videos with an API
Send ReelSDK the episode audio and the cover image, and in under 30 seconds you have a 1080x1920 video with a moving waveform; one more request burns in the subtitles.
Use casesUpdated
The request
POST /v1/ffmpeg
Turn a podcast clip and cover art into a vertical waveform video with burned-in captions, using two API calls.
curl "https://api.reelsdk.com/v1/ffmpeg?wait=300" \
-H "Authorization: Bearer $REELSDK_KEY" \
--json '{
"inputs": {
"episode.mp3": "https://api.reelsdk.com/f/lib/in/nasa-sciencecasts/coffee-cup-30s.mp3",
"cover.jpg": "https://api.reelsdk.com/f/lib/in/nasa-sciencecasts/coffee-cup-cover.jpg"
},
"args": [
"-loop", "1", "-framerate", "30", "-i", "cover.jpg", "-i",
"episode.mp3", "-filter_complex",
"[0:v]split[c1][c2];[c1]scale=1080:1920:force_original_aspect_ratio=increase,crop=1080:1920,gblur=sigma=40,eq=brightness=-0.32:saturation=1.3[bg];[c2]scale=960:540[art];[1:a]showwaves=s=960x240:mode=cline:rate=30:colors=0xFFD166:scale=sqrt:draw=full[wave];[bg][art]overlay=60:420[v1];[v1]drawbox=x=60:y=1129:w=960:h=2:color=0xFFD166@0.5:t=fill[v1b];[v1b][wave]overlay=60:1010:shortest=1,drawtext=fontfile=/usr/share/fonts/reelsdk/Montserrat-ExtraBold.ttf:text='\''THE ZERO GRAVITY COFFEE CUP'\'':fontsize=50:fontcolor=white:x=(w-text_w)/2:y=300,format=yuv420p[v]",
"-map", "[v]", "-map", "1:a", "-c:v", "libx264", "-preset", "medium",
"-crf", "21", "-c:a", "aac", "-b:a", "128k", "-shortest", "-movflags",
"+faststart", "audiogram.mp4"
],
"outputs": ["audiogram.mp4"]
}'const res = await fetch("https://api.reelsdk.com/v1/ffmpeg?wait=300", {
method: "POST",
headers: { Authorization: `Bearer ${process.env.REELSDK_KEY}`, "Content-Type": "application/json" },
body: JSON.stringify({
"inputs": {
"episode.mp3": "https://api.reelsdk.com/f/lib/in/nasa-sciencecasts/coffee-cup-30s.mp3",
"cover.jpg": "https://api.reelsdk.com/f/lib/in/nasa-sciencecasts/coffee-cup-cover.jpg"
},
"args": [
"-loop", "1", "-framerate", "30", "-i", "cover.jpg", "-i",
"episode.mp3", "-filter_complex",
"[0:v]split[c1][c2];[c1]scale=1080:1920:force_original_aspect_ratio=increase,crop=1080:1920,gblur=sigma=40,eq=brightness=-0.32:saturation=1.3[bg];[c2]scale=960:540[art];[1:a]showwaves=s=960x240:mode=cline:rate=30:colors=0xFFD166:scale=sqrt:draw=full[wave];[bg][art]overlay=60:420[v1];[v1]drawbox=x=60:y=1129:w=960:h=2:color=0xFFD166@0.5:t=fill[v1b];[v1b][wave]overlay=60:1010:shortest=1,drawtext=fontfile=/usr/share/fonts/reelsdk/Montserrat-ExtraBold.ttf:text='THE ZERO GRAVITY COFFEE CUP':fontsize=50:fontcolor=white:x=(w-text_w)/2:y=300,format=yuv420p[v]",
"-map", "[v]", "-map", "1:a", "-c:v", "libx264", "-preset", "medium",
"-crf", "21", "-c:a", "aac", "-b:a", "128k", "-shortest", "-movflags",
"+faststart", "audiogram.mp4"
],
"outputs": ["audiogram.mp4"]
}),
});
const { job } = await res.json();{
"job": {
"id": "job_iczp2zbxya2k15xklbgx",
"op": "ffmpeg",
"status": "succeeded",
"stage": null,
"created_at": "2026-11-07T03:15:37.227Z",
"finished_at": "2026-11-07T03:16:00.732Z",
"outputs": {
"audiogram.mp4": "https://api.reelsdk.com/f/out/job_iczp2zbxya2k15xklbgx/audiogram.mp4?s=..."
},
"result": null,
"duration_s": 30,
"error": null,
"usage": {
"output_seconds": 30,
"audio_seconds": 0,
"tts_characters": 0,
"images": 0
},
"poll_url": "https://api.reelsdk.com/v1/jobs/job_iczp2zbxya2k15xklbgx"
}
}
Call 1: cover, title and waveform
POST /v1/ffmpeg runs ffmpeg 8.1 with your own filter graph. For this clip, the graph loops the cover image, blurs a copy for the background and puts the cover on top at 960x540. Then it writes the title with drawtext and draws the audio with showwaves at 960x240, in cline mode on a sqrt scale.
If you want a different look, showwaves also has point, line and p2p modes alongside lin, log and cbrt scales, and the allowlist has more audio visualisation filters than the 10 in the table, such as abitscope, aphasemeter and showspatial. This first pass writes a 1080x1920 video at 30 fps, as long as the audio, and it came back in about 24 seconds.
At a glance
Audio visualisation filters the ffmpeg endpoint accepts
All 10 are on the allowlist. Descriptions from the FFmpeg filter docs.
| Filter | What it draws | Output |
|---|---|---|
showwaves | Convert input audio to a video output, representing the samples waves. | Video, this run |
showwavespic | Convert input audio to a single video frame, representing the samples waves. | One image |
showspectrum | Convert input audio to a video output, representing the audio frequency spectrum. | Video |
showspectrumpic | Convert input audio to a single video frame, representing the audio frequency spectrum. | One image |
showcqt | Frequency spectrum on a musical tone scale, with a constant Q transform. | Video |
showcwt | Frequency spectrum using Continuous Wavelet Transform and Morlet wavelet. | Video |
showfreqs | Audio power spectrum: amplitude on the Y axis, frequency on the X axis. | Video |
showvolume | Convert input audio volume to a video output. | Video |
avectorscope | Convert input audio to a video output, representing the audio vector scope. | Video |
ahistogram | Convert input audio to a video output, displaying the volume histogram. | Video |
Sources: FFmpeg filters documentation, ReelSDK ffmpeg docs, checked November 2026.
Call 2: captions from the audio
POST /v1/captions takes the finished audiogram URL from the first call, transcribes the speech, and burns in the words. This run uses karaoke: Poppins at the bottom, 5 words per group, with the spoken word highlighted as the clip plays.
Along with the captioned video, the endpoint returns captions.srt, so you can send the same clip to a player that uses closed captions. That call took about 17 seconds and cost 1.5 cents, since captions cost $0.03 per output minute. See captions docs.
curl "https://api.reelsdk.com/v1/captions?wait=300" \
-H "Authorization: Bearer $REELSDK_KEY" \
--json '{
"video_url": "https://api.reelsdk.com/f/lib/out/job_iczp2zbxya2k15xklbgx/audiogram.mp4",
"style": "karaoke",
"language": "en"
}'What it costs per clip
ffmpeg is billed at $0.02 per compute minute, per second of the time spent fetching inputs, running ffmpeg and storing outputs. The first call billed about 20 seconds of compute, which is about two thirds of a cent. Add the captions pass and the whole clip came to about 2.2 cents.
If you pass arguments outside the allowlist, you get a 400 INVALID_REQUEST answer right away before any job starts. You can read the ffmpeg docs and check pricing for the rest of the details.
What it costs
What this audiogram cost
- Call 1, ffmpeg, 19.7 s of compute at $0.02 per minute
- $0.0066
- Call 2, captions, 30 s at $0.03 per minute
- $0.015
- Total
- $0.0216
ffmpeg is billed per second of the time spent fetching inputs, running ffmpeg and storing outputs, with a 1 s minimum. Machine start-up is not billed, and a failed job costs nothing.
Limits
What one request takes
- Inputs
- Up to 50 files
- Outputs
- 1 to 20 files
- Timeout
- timeout_s from 1 to 600
- Frame size
- Up to 8192 px a side and 37.7 megapixels
- Input files
- 500 MB each, 1 GB per job
- Media length
- Up to 15 minutes for captions and slideshows
- Job run time
- Up to 10 minutes
- Concurrent jobs
- 3 per key
- Jobs per day
- 500 per key
- Output retention
- 7 days
How it works
From request to file
Host the clip and the cover
Put the audio and cover image at public URLs, or upload them with
POST /v1/uploads.Send POST /v1/ffmpeg
List both files in
inputs, pass your filter graph inargs, and name the output inoutputs.Send POST /v1/captions
Pass the audiogram URL from call 1 as
video_urlalong with a caption style.Save the files
Download the captioned video and
captions.srtbefore the links expire in 7 days.
FAQ
Questions, answered
What does an audiogram cost?
About 2.2 cents for a 30-second clip: about two thirds of a cent for the ffmpeg call and 1.5 cents for the captions call.
Can I change the waveform style?
Yes, showwaves takes point, line, p2p or cline modes, a color, a size and a lin, log, sqrt or cbrt scale. More than 10 audio visualisation filters are allowed, from spectrums to vector scopes.
Do I need captions?
No, the first call alone gives you the waveform video. The captions call adds burned-in words and an SRT file.
Can I use my own font for the title?
Yes. Add the font file to inputs by public URL (POST /v1/uploads doesn't take font files) and point drawtext at it, or use one of the bundled fonts.
How long can the clip be?
The ffmpeg step doesn't check input length and runs up to 600 s, set with timeout_s. The captions pass takes video up to 15 minutes.
- Video to SRT APISend a video URL and get SRT and VTT subtitle files, a word-timed transcript and a captioned copy of the video back.
- Quote videosRender a quote video from text, a background photo and a music track with one ffmpeg call: slow zoom, fade and centred type.
- FFmpeg API: run ffmpeg commands over HTTP, $0.02 a compute minuteRun ffmpeg commands over HTTP: send input URLs and your arguments, get output URLs back. ffmpeg 8.1 on a fresh 4 vCPU machine per job, ready in about 2 s, $0.02 per compute minute.
- FFmpeg API reference: POST /v1/ffmpegPOST /v1/ffmpeg runs ffmpeg 8.1 with your own arguments: up to 50 inputs of 500 MB, 20 outputs and 600 s per command, with tested recipes for GIFs, thumbnails and subtitles.
- Captions API reference: POST /v1/captionsBurn animated word-by-word captions into videos up to 15 minutes with POST /v1/captions: 3 presets, 4 fonts, auto language detection, and SRT, VTT and transcript files.
Ready to build?
Ship video from your app today
Your first $0.50 is free, and one key covers every endpoint.
