Initial Audio Rendering missing samples....

[belisoful]

Apple Feedback Assistant ID: MISSING

DESCRIBE THE BUG:

I have an Storyline with audio only. It needs to seemlessly repeat an audio sample many times. I have 1 frame of silence to set the zero point. The audio renders differently if that first frame is turn on or off, when it shouldn't.
This glitch in exporting audio happens when rendering video with it as well. Having a built in fade doesn't help and is affected by this. Manual fading is affected by this too.

---

TO REPRODUCE:
In Audacity, Generate a "silence" Flac file of 1 second.
IIn Audacity Generate a "tone" Flac file of a minute or ten. It just needs to be constant sound for the whole file.
In FCPX, Drop in the silence and make it 1 frame long and "Create a Storyline" on it.
Drop in the tone file into the storyline after the silence frame.
Export the Audio only as WAV.

Turn off the 1 frame of silence.
Export the Audio only as WAV again.

Compare the two audio exports in Audacity.

---

EXPECTED BEHAVIOUR:
Turning on and off the 1 frame of silence should not have any effect on the test tone audio rendering.

---

SCREENSHOTS:
Here is what it looks like with the 1 frame of silence. this is correct.
And here is what it looks like with the 1 frame of silence turned off. This produces an audible click, and is an error.
Here are the combined images scaled properly.

there are missing samples at the very beginning.
The silence and audio test tone was 96khz- 24bit flac.
The project is 48 kHz audio, and 24 FPS. I've seen this audio bug with other sample rates.
As you can see, the silence is mono. the 40 hz Harmonics is stereo.
and there is nothing extending or strange with the Audio Component/expansion. Everything is lined up correctly.

---

SPECS:

• 2021 16-inch MacBook Pro (M1 Max, 32GB RAM, 2TB SSD)
• macOS Sonoma 14.6.1
• Final Cut Pro 10.8.1

---

ADDITIONAL COMMENTS:

I have extensively run tests on the audio rendering algorithm testing the test tone in a storyline without the 1 frame of silence. All the audio export algorithms were tested, here are the results.

🪳 AIFF has file size limitations that are not warned about. If you try rendering a file larger than it can handle, it will try but get cut off. No warning. All the stuff the file can't handle (due to file size limitations) is rendered and promptly ignored.

WAV and CAF holds a properly rendered file. The test tone shows up in the very first sample without missing samples, and is the exact length it should be to the sample.

AAC, AC3, and MP3 had weird things going on with being too short, or slightly longer than the total video length. They also added silence at the beginning (~9ms, ~9ms, and ~43ms, respectively), but I think that's the format rather than the audio rendering. This needs confirmation though. My initial tests were rudimentary.

The fact that there was several frames of audio missing at the end in some lossy formats was disturbing.

Apple has not VALIDATED their audio rendering for lossy encoding formats. I suspect it's a regression issue.... they didn't do regression they should have done.

[belisoful]

I'm testing this in FCPXI right now.

..... or is it FCP11? Or are we just retaining FCPX?

[belisoful]

This is 100% still a bug in FCP11. I tested it with 60 fps. The tone started at exactly 1 frame after the silence as expected.

The silence being turned off auto-destructed those exact same ~5ms of audio in the tone, even with a new 60 fps project. So, the bug is Frame rate agnostic.