Skip to content

Your first audio-reactive visual Beginner

Objective — build a visual that reacts to sound coming into your mic, with one live control you can perform, entirely by asking in plain language.

What you'll see — a glowing spectrum that rises and falls with the music, plus a placeholder frame whose brightness pulses to the beat. Turn one knob and the whole scene gets more or less sensitive.

The reactive output frame pulsing in colour and brightness with the sound (captured with a test tone).

Before you start

  • tdmcp installed for your AI assistant (Claude or Codex).
  • TouchDesigner open, with bridge running in the Textport. See Install for Claude if you don't.
  • A microphone. No mic? No problem — a step below uses a test tone instead.

Steps

Type each prompt to your AI assistant, one at a time. Wait for it to finish before sending the next.

  1. Check TouchDesigner is connected.

    text
    Check that TouchDesigner is connected and tell me what version I'm on.

    The assistant runs a quick health check (get_td_info). You should get back a version and "connected". If not, see If it goes wrong.

  2. Build the audio-reactive starter.

    text
    Apply the audio_reactive_basic recipe. If my mic isn't available, use a test tone
    instead so I can still see it react.

    This drops in a validated network: an audio input feeds a spectrum analyzer and an RMS level, and leaves a placeholder frame ready to react.

  3. Make the picture react to the sound.

    text
    Bind the placeholder frame's brightness to the RMS level so it pulses with the music.

    The assistant wires the audio level (the level_null channel) into the frame's color using bind_to_channel. Play some music or speak — the frame should pulse.

  4. Add a spectrum bar visual.

    text
    Add an audio_spectrum_bars visual next to it, driven by the same audio, so I can
    see the frequencies as colored bars.

    You get the classic analyzer look: a row of glowing cyan-to-magenta bars rising and falling with the music.

  5. Expose one control you can perform.

    text
    Expose a single Sensitivity control I can turn live to make the whole scene react
    more or less strongly.

    The recipe's built-in Sensitivity control (range 0–4) becomes a live slider. Turn it up and quiet sounds pop; turn it down to calm it.

  6. See it.

    text
    Auto-arrange everything and show me a preview of the output.

    The assistant lays the network out left-to-right, checks for errors, and returns a thumbnail of the result.

Expected result

In TouchDesigner you'll see a tidy left-to-right chain: audio in → spectrum + RMS level → a spectrum-bars visual and a pulsing placeholder frame → a final output. The preview shows glowing bars and a frame that breathes with the sound, and dragging the Sensitivity slider visibly changes how hard everything reacts.

If it goes wrong

  • Nothing reacts / it's silent — your mic may be muted or unavailable. Ask: "Switch the audio source to a test tone." Then confirm your OS lets TouchDesigner use the microphone. More in Troubleshooting.
  • No preview appears / "can't reach TouchDesigner" — the bridge isn't running. Check the Textport for bridge running and re-run step 1. See Troubleshooting.
  • It reacts too weakly or too strongly — turn the Sensitivity control from step 5, or ask "Raise the Sensitivity default."
  • Still stuck? — the FAQ covers the most common first-run questions.