Audiogram maker

Make a podcast audiogram in five steps

1. Choose your episode or a clip of it (MP3, WAV, M4A and other common formats), or load the short original sample.
2. Type the start and end of the moment you want; the first 60 seconds are selected to begin with. Clips of 30 to 90 seconds work best on social feeds, and the page reminds you when a clip is shorter or longer.
3. Add your cover art, the episode title and the show name. The shape starts as vertical 9:16 for Reels, TikTok and Shorts; switch to square for Instagram and LinkedIn feeds.
4. Press Generate captions to transcribe the clip on your device, or paste or load your own SRT file. Fix any wrong words in the caption box; the preview updates as you type.
5. Press play to check the preview, then Export video and download the MP4.

What makes an audiogram work

Most people scroll social feeds with the sound off, so a podcast clip has to show what is being said. An audiogram does that with three things: captions that carry the words, a moving waveform that shows the clip is audio, and your artwork and title so people know which show it is. Pick a moment that makes sense on its own, starts with a strong first sentence and ends on a point rather than mid-thought. Thirty to ninety seconds is a good length for most platforms; very short clips rarely carry an idea, and long ones lose viewers.

How the captions are made

When you press Generate captions, your browser downloads Whisper base.en, an open speech-recognition model from OpenAI (MIT licence), and the engine that runs it, about 107 MB in total, then transcribes only the clip you selected. Nothing is downloaded before you press the button, and the audio never leaves your device. The words come back with timings and are grouped into short caption lines of up to about 32 characters, so they stay readable on a phone. The result is plain SRT text in a box you can edit, so you can fix a name or remove a word before export. If you already have a transcript from your editing software, load its SRT file instead: the times should count from the start of the original file you chose here.

Honest limits

This is a manual clip maker. It does not listen to a whole episode and pick viral moments for you, the way AI clipping services such as Opus Clip do, and it does not reframe video or track faces: you choose the moment, and it makes a clean, captioned audiogram from it, privately and free. Caption accuracy depends on the recording: clear speech in English works well, overlapping speakers, heavy accents and music beds make more mistakes, and the model does not tell speakers apart. Exports run on your own device, so a phone is slower than a laptop, and one video can be up to 10 minutes long.

Format and privacy

The video is an MP4 with H.264 video and AAC audio where your browser can encode them, at 1080 × 1080, 1080 × 1920 or 1920 × 1080, at 30 or 60 frames per second. Browsers without an H.264 encoder get a WebM file instead, and the page tells you. Every frame is drawn and encoded inside this browser tab, and your episode, artwork and captions are never uploaded to AudioClover.