A voice agent waveform that cycles through the four states a voice AI is ever in: idle, listening, thinking and speaking. For voice AI, AI receptionist, AI call centre and AI SDR sites that would otherwise paste in a video of a waveform.
Each state has its own motion. Idle settles to a low line with a slow breath. Listening rises through the middle and jitters at a sensitivity you set. Thinking runs a single ripple left to right. Speaking is a full waveform with syllable-like bursts. States crossfade rather than jumping.
There is no microphone in it. It never asks a visitor for audio permission and there is no audio API in the file at all: every bar height is arithmetic over elapsed time, so it behaves identically for every visitor and asks them for nothing.
Leave it on auto and it cycles, with its own hold duration per state, or pin one state so you can screenshot each. Bar count, width, gap and radius, wave height, every colour, the four state labels (so they can be translated) and the font are property controls.
The bars are written straight onto their own nodes as a transform rather than re-rendered every frame. In the Framer canvas, and for a visitor who has asked for reduced motion, it draws a full speaking waveform rather than a flat line.
After purchase you receive the component URL. In Framer, press Insert, choose Add Component URL, and paste it.
Licence: yours to use on your own sites and on client sites, commercially, with no attribution. Not for resale or redistribution: do not share the component URL, and do not publish the component or its code anywhere, paid or free. Full terms arrive with your purchase. Questions: me@majedesigns.dev
A voice agent waveform that cycles through the four states a voice AI is ever in: idle, listening, thinking and speaking. For voice AI, AI receptionist, AI call centre and AI SDR sites that would otherwise paste in a video of a waveform.
Each state has its own motion. Idle settles to a low line with a slow breath. Listening rises through the middle and jitters at a sensitivity you set. Thinking runs a single ripple left to right. Speaking is a full waveform with syllable-like bursts. States crossfade rather than jumping.
There is no microphone in it. It never asks a visitor for audio permission and there is no audio API in the file at all: every bar height is arithmetic over elapsed time, so it behaves identically for every visitor and asks them for nothing.
Leave it on auto and it cycles, with its own hold duration per state, or pin one state so you can screenshot each. Bar count, width, gap and radius, wave height, every colour, the four state labels (so they can be translated) and the font are property controls.
The bars are written straight onto their own nodes as a transform rather than re-rendered every frame. In the Framer canvas, and for a visitor who has asked for reduced motion, it draws a full speaking waveform rather than a flat line.
After purchase you receive the component URL. In Framer, press Insert, choose Add Component URL, and paste it.
Licence: yours to use on your own sites and on client sites, commercially, with no attribution. Not for resale or redistribution: do not share the component URL, and do not publish the component or its code anywhere, paid or free. Full terms arrive with your purchase. Questions: me@majedesigns.dev
