Video volume increases abruptly at the end + audio mixing cracks

[yuvraj108c]

I have rendered the following video for a project using remotion. It consists of a background music, alongside with some voice overs played at different time frames.

video.mp4

• If you listen carefully, there is an audio crack when the second voiceover (audio_2) starts playing. The same problem occurs if I add audio_3, audio_4 etc
• At the end of the video, the volume increases unexpectedly.

The video works fine during the preview mode. I believe the video render process is causing this weird bug. I have crosschecked the audio files and they are all of good quality, with no cracks/noises. This is my code below.

import bgMusic from "./assets/bgMusic.mp3";
import audio_1 from "./assets/1.mp3";
import audio_2 from "./assets/2.mp3";
import { Audio, Sequence } from "remotion";

export const Bug: React.FC<{}> = () => {
  return (
    <div className="audios">
      <Audio src={bgMusic} volume={0.3} />;
      <Sequence from={15}>
        <Audio src={audio_1} />
      </Sequence>
      <Sequence from={90}>
        <Audio src={audio_2} />
      </Sequence>
    </div>
  );
};

[JonnyBurger]

Hi @yuvraj108c! Thanks for the report.
I can understand the issue arising due to how FFMPEG works.

To investigate, I need the 3 MP3 files, can you share them as well? If you want, privately under jonny@remotion.dev - I'll take a look at what we can do to improve the mixing in your scenario!

[yuvraj108c]

Thanks for replying. I have zipped the 3 audios here: audios.zip

[JonnyBurger]

Hey Yuvraj,

thanks a lot for uploading! I triggered an investigation and we can confirm the bug. After some trying out, I found a sequence of commands that will generate a better audio than our current mechanism:

ffmpeg -i bgMusic.mp3   -ac 2 -af "[0:a]atrim=0.000:5.000,volume=0.3:eval=once[a0]" out/bgMusic.mp3    -y
ffmpeg -i 1.mp3         -ac 2 -af "[0:a]atrim=0.000:5.000,adelay=500|500|500,atempo=1.00000,apad=whole_dur=5.000,volume=1:eval=once[a0]"  out/1.mp3          -y
ffmpeg -i 2.mp3         -ac 2 -af "[0:a]atrim=0.000:5.000,adelay=3000|3000|3000,atempo=1.00000,apad=whole_dur=5.000,volume=1:eval=once[a0]"       out/2.mp3          -y
ffmpeg -i out/bgMusic.mp3 -i out/1.mp3 -i out/2.mp3 -filter_complex "[0:a][1:a]amerge=inputs=3,pan=stereo|c0<c0+c2+c4|c1<c1+c3+c5[a]" -map "[a]" out/final.mp3        -y

We will refactor our audio pipeline and make it better (also thanks to your report), but we have to prioritize the 3.0 launch that we've been working on for some time now. Sorry for this problem and that we cannot take prompt action to fix it 😌 An alternative for now could be to mix the audio yourself using for example the FFMPEG way showed above.

[yuvraj108c]

No worries Jonny, I will use FFMPEG for now until the issue is resolved.

Looking forward for the v3 launch.