You can make a difference in the Apple Support Community!

When you sign up with your Apple Account, you can provide valuable feedback to other community members by upvoting helpful replies and User Tips.

Siri-esque audio visualization?

Can anyone recommend a plugin or site that can generate animated visualization of speech that looks a bit like the Siri one? We have a scene where someone is conversing with a laptop. Thanks for any pointers.



Posted on Sep 23, 2024 5:04 PM

Reply
19 replies

Sep 28, 2024 2:56 PM in response to terryb

Here's the download:

https://fcpxtemplates.com/sdm_downloads/siri-animation-effect/


Here's a working demo:

https://youtu.be/t7x_oSTu6xQ


What's in the download:

  • Siri Animation (the Effect for Final Cut Pro)
  • Clone Layer (most think of this as an "adjustment layer" - adaptable for variable lengths [more below])
  • The original development project "siri-animation-3c" (This is a Motion Project/Composition for Ben to use to hook up audio to the animation — there are no published parameters in this.)
  • License


Compromises were made. The Effect and the Project are 10 minutes long. A Build In Mandatory is used to maintain timing, so any application longer than 10 minutes will simply stop. The Effect cannot be "wired" to any audio - the Effect is "faked" by a randomizing setup.



There is a built in "In Animation" which can be varied from NO animation to 2 seconds.


The Animation is based on Start Point and End Point (and there are OSCs for you to position these). The OSCs will adapt to any media format that the Effect is applied to. You can move these locations anywhere you like, but animating only the End Point will be meaningful. The Start Point is useless after the first frame is played.


Highlight Visuals (the "flashing parts")

This is how it works: Whenever the parameter is an Integer value, there are no flashes. Whenever the value is a decimal/fractional, flashing occurs. You create timed flashing by **keyframing** the value over time from one integer value to a higher (or lower) one. When the frames are played, the parameter will pass through the decimal values and depending on how far you increment the value will determine the number of "beats" that occur. However, the effect is *further* randomized so that "beating" will become obscured and the "waveform" looks more "natural". Randomization incursion is influenced by the Frequency and how it is applied is also influenced by Random Seed. It is recommended to keep number spans and Frequencies less than 10 and experiment from there. Remember to occasional hit Random Seed to get better "audio matches".


As your video progresses, and you are "talking to Siri", you will need to set a Keyframe at the beginning of the flashing parts, move the playhead to the end of flashing and **choose a number to progress to**. You can continue this on for up to 10 minutes of "conversation" with Siri.


The Rest:

There is an optional Solid Background option.

A Color can be chosen and its Opacity can also be selected/keyframed as part of the overall effect (as in the video with the fade out at the end.)


You may set the Siri Size for either the whole length of the effect (except for In Animation which will start at 0), or animate this over the "life" of Siri.


Drop Shadow Opac. and Drop Shadow Dist. can be adjusted to taste. When animating Siri Size, you may also want to animate the Distance.


Show OSCs : OSCs are off by default, but can be turned on to maneuver Siri for In Animation setup.



The FCP Effect is locked, you will not be able to open it in Motion — it contains techniques which I prefer to keep private for now. The Composition is "open" so that the actual "artwork" is available.


Remember that your setups for the FCP Effect can be saved as Effects Presets. Save your setups *without* Highlight Visuals keyframing... you'll want to handle that on an individual application basis.


HTH



Sep 26, 2024 9:56 PM in response to terryb

Um... Audio response for a final cut template is not possible, but it is possible in Motion IF you import the audio, clear the "old" audio from the behaviors and hook up the new audio — it's a royal pita. It can be faked in FCP.


There are a lot of aspects to this. Animating on/off. As you talk, it gets larger and when it's idle, it reduces to about half size - it doesn't simply dismiss itself...


Bear with me. There are a lot of subtleties to work out.


A couple more screenshots:


"vocalization variations"




(still needs some work...)



Sep 27, 2024 10:22 AM in response to fox_m

Thanks again Fox for diving so deep into this. It's a bummer there isn't a direct way to use audio amplitude to drive an effect parameter. ffmpeg can export a csv of timestamps and RMS dB values for a given audio file (https://stackoverflow.com/questions/38056970/ffmpeg-txt-from-audio-levels). Could a plugin use this output as input to drive animation?

Sep 28, 2024 5:13 PM in response to BenB

@BenB:


How to hook up the composition with audio.


1 Import your Audio clip

2 Locate "Link2Osc" in Group 2

3 disconnect all the behaviors (Randomize, Clamp, Link and Oscillate)

4 Right click on the Numbers > Value parameter and add an Audio Behavior

5 Drop the audio in the drop zone and let it "cook"

... Play ("Time" the audio clip — trim the Play Range)


HTH

Siri-esque audio visualization?

Welcome to Apple Support Community
A forum where Apple customers help each other with their products. Get started with your Apple Account.