Cadence

Cadence supports After Effects CC 2018 or later, and its listed compatibility runs through CC 2018, CC 2019, 2020, 2021, 2022, 2023, 2024, 2025, and 2026. If you're on a recent release, the plugin is covered. Because the speech-to-marker step runs through an AI service rather than locally, you'll also need an active internet connection and a Replicate account for processing. Check the developer's page for the latest compatibility details before installing on a specific build.

Cadence requires an internet connection because its AI processing runs remotely rather than on your machine. You'll need a Replicate account, set up through replicate.com, to handle marker generation. Once connected, you select an audio layer and click Generate word markers, and Cadence uses WhisperX to place a frame-accurate marker on every spoken word. Without a connection or a configured API account, the AI step can't run, so plan for online access whenever you process audio.

Cadence supports 90+ languages with frame-accurate precision, including English, French, Spanish, German, and Japanese among others. Under the hood, Wav2Vec2 detects exact word boundaries with phoneme-level precision, and word-level forced alignment places each marker at the exact millisecond a word begins. That accuracy holds across long files too, with zero drift keeping sync perfect from the first word to the last, even on hour-long audio. For the full language list, see the developer's page.

Cadence runs on Windows 10 and 11 as well as macOS, so both major desktop platforms are covered. It works inside After Effects CC 2018 or later, and because the AI processing happens remotely, an internet connection and a Replicate account are also needed. After setup, you select one or more audio layers and generate word markers in a single operation, with each marker landing on the audio layer non-destructively so your comp stays clean.

Cadence is listed as free in this catalog, so there's no cost noted here to start using it. Licensing terms can change, though, so check the developer's page for the current details before you download. Functionally, it places a frame-accurate marker on every spoken word in your selected audio, labels each marker with the word, and lets you copy those markers to the current timeline, parent layer, or parent timeline with one click.