Audio for YouTube uploads

Howard-Vigorita wrote on 9/25/2025, 10:12 PM

Been trying to get some feedback on possibly getting better quality audio onto and out of YouTube. But so far not much response. I'm thinking maybe no one really expects good audio from YouTube and/or doesn't care. Anyway, thought I'd take a shot at getting some responses here.

I uploaded a couple of short 4K test clips to YouTube a couple of days ago. One of them is the mp4/aac format YouTube recommends. I don't really think much of aac audio quality which is why I'm doing the test. The other clip is in mov/wav format which obviously sounds better to me and mine if I play it here at home through my JBL powered studio monitors or the passive JBL's in my living room. But the 64 thousand dollar question is whether YouTube listeners can possibly hear any difference after YouTube converts my video clips to vp9 with opus-audio soundtracks. Which is supposed to sound better than aac. If opus really is better sounding than aac, listeners with decent headphones or earbuds should hear some difference. So far I've gotten only 2 on-point responses in 2 days. And only 1 thinks he hears a difference. And that one is a big maybe.

The mp4 clip I uploaded to YouTube was made by Vegas 23 using a Sony wav64 24bit 48k audio mixdown also made by Vegas. The mov clip was made by rendering the audio track from the same project in Microsoft 24bit 48k WAV format which was then swapped for the aac audio stream by ffmpeg while repackaging as a MOV clip. The 4k video streams were made by MainConcept set for 28mbps cbr... but that shouldn't have any impact on the audio stream. The mics and recorder I used to record the audio are among the highest rated recording studio quality: DPA's on piano. And a custom-built tube mic (Bock 151) for the audio overdub through a vintage-design Neve preamp. The Neuman kms105 stage mic seen in the video was replaced by the same singer's studio overdub. Reverb is a state of the art hardware unit: a Bricasti M7... I happened to have collaborated with one of the designers years ago (Casey, who is the cas in Bri-cas-ti) on the presets and their manipulation for desired spacial psycho-acoustic effects.

Anyway, here's a link to the 1st test clip, whose description contains a link to the 2nd one:
www.youtube.com/watch?v=JxmtDSPk1Uo

Any comments, here or in the YouTube comments, would be greatly appreciated.

Comments

EricLNZ wrote on 9/25/2025, 11:56 PM

Watching on my pc I cannot tell any difference but I'm not musically tuned.

Alan-Smithee wrote on 9/26/2025, 9:03 PM

Other possibility is uploading in 5.1 , either left and right channels only active the others mute, or left and right channel as normal with front, surround rears at -6db, with LFE mute. I have not done either

When I listen to the 5.1 music on a phone using it's speaker or with headphones the fidelity of the music is better with the 5.1 track compared to the stereo opus tracks. It's not an apples to apples comparison though - for YT to serve the dolby digital encode must turn on the phone's surround mode, and for opus the surround has to be turned off.

There's a further potential problem with the YT downmix - possibly a L + R only 5.1 will sound the same on the opus stereo track, but a 5track version could sound worse. There's another problem I've heard about where Dolby Digital audio is normalized at a lower volume and may affect the opus track. Unless you're listening through a better than average phone or your TV's surround speakers only the opus track will be heard so that's got to sound good.

An example where the music seems to have better fidelity with the 5.1 track

Alan-Smithee wrote on 9/26/2025, 9:35 PM

(Correction 5.1 audio from YT I've compared is Dolby Digital Plus (E-AC-3)

Howard-Vigorita wrote on 10/9/2025, 10:43 PM

Been having no issues with doing my uploads to YouTube of mp4's rendered from Vegas repackaged into mov containers with 24bit WAV audio. And got a few positive feedback comments on the audio quality so I guess it's not lost on everyone. I just updated the powershell/ffmpeg script I've been using to enable fast-start for a little quicker YouTube processing:

# sweep through a folder of mp4 clips replacing aac audio streams with wav streams in mov containers
# place this script in the folder containing mp4 clips
# place replacement audio in WAV folder with matching BaseNames
# script will generate mov clips in a pre-existing MOV folder
# tested sucessfully under ffmpeg v7.1.1 with pcm clips in Microsoft 24bit wav format
# updated to enable faststart for quicker YouTube processing

$files = Get-ChildItem -Path "." -Filter "*.mp4"
measure-command
 {
    foreach ($file in $files)
      {
         $IN=$file.BaseName+'.mp4'
         $AUDIO='.\wav\'+$file.BaseName+'.wav'
         $OUT=$file.BaseName+'.mov'

        # swap audio
        ffmpeg -i "$IN" -i "$AUDIO" -movflags +faststart -map 0:v:0 -map 1:a:0 -c:v copy -c:a copy -y "mov\$OUT"

    }
}
pause

Howard-Vigorita wrote on 10/10/2025, 11:18 AM

Btw, the measure-command function is powershell's built-in timer. This script ripped through 18 mp4 clips on my 11900k/5080 system that were rendered nvenc hevc vbr 35/70 mbps by Vegas 23 and turned them all into mov/wav clips in 40 seconds. The matching mp4 and wav clips were all made with the Batch Render Timer script linked in my signature.