Mixing audio for YouTube, DVD and MP4 File Playback

discburn wrote on 12/6/2018, 7:55 AM

I’ve always mixed DVDs at Peak of -6dB to avoid distortion in TVs (even Flat Screens).

And I do mix loud for YouTube, leaving headroom of -.5dB.

What to do, though, with 'video files' (MP4) that will play into TVs from USB flash drives, or from DVD players with hard drives and HDMI out? If the DVDs cause distortion, won’t the MP4s do the same unless peaking at -6dB.

And yet, most material nowadays seems to be loud (approx. CD level, near 0dB). Are modern speakers in TVs better at handling higher levels?

Seamus

Comments

CubeAce wrote on 12/9/2018, 4:49 AM

@discburn

Every 3dB you reduce, you are halving your volume. A reduction of just -1dBm should be sufficient to stop clipping. Using some sort of hard limiter will stop the volume from reaching that, but audio mixing is an art in itself and relies on the quality at every stage being as good as it can be. Videos you hear without distortion are probably using high end components from the microphone and pre-amp stage through to the final mixdown and using some form of multiband compression/expansion system in both the initial recording and final mixdown. Any sharp transient that doesn't reach a natural peak before being hit with a limiter will sound distorted whatever level you reproduce it at as it un-naturally flattens the peak at whatever level it is set at, and anything above that set level will flatten abruptly and sound distorted. Metering monitoring systems and automation in the sound levels in recording can be too slow to stop that happening or too fast for your eye to register. TV speakers are not the best nor are the amplifiers that feed them, but the pre-amps often are ok to feed into a better system. Set source recording levels manually and leave some room for peak levels that the meters may be too slow to react to. If you have the facility to do so, add a light compression to the source at the time of recording. If you have peak hold on your meters, use it to find the peaks and lower the mix volumes accordingly. Do not go too far the other way either. There are some decent books on audio recording including 'The sound engineers handbook', that may seem like overkill but are really useful in getting to grips with audio recording.

Last changed by CubeAce on 12/9/2018, 4:52 AM, changed a total of 1 times.

 

Windows 10 Enterprise. Version 22H2 OS build 19045.5737

Direct X 12.1 latest hardware updates for Western Digital hard drives.

Asus ROG STRIX Z390-F Gaming motherboard Rev 1.xx with Supreme FX inboard audio using the S1220A code. Driver No 6.0.8960.1 Bios version 1401

Intel i9900K Coffee Lake 3.6 to 5.1GHz CPU with Intel UHD 630 Graphics .Driver version Graphics Driver 31.0.101.2135 for 7th-10th Gen Intel® with 64GB of 3200MHz Corsair DDR4 ram.

1000 watt EVGA modular power supply.

1 x 250GB Evo 970 NVMe: drive for C: drive backup 1 x 1TB Sabrent NVMe drive for Operating System / Programs only. 1X WD BLACK 1TB internal SATA 7,200rpm hard drives.1 for internal projects, 1 for Library clips/sounds/music/stills./backup of working projects. 1x500GB SSD current project only drive, 2x WD RED 2TB drives for latest footage storage. Total 31TB of 10 external WD drives for backup.

ASUS NVIDIA GeForce RTX 3060 12GB. nVidia Studio driver version 572.60 - 3584xCUDA cores Direct X 12.1. Memory interface 192bit Memory bandwidth 360.05GB/s 12GB of dedicated GDDR6 video memory, shared system memory 16307MB PCi Express x8 Gen3. Two Samsung 27" LED SA350 monitors with 5000000:1 contrast ratios at 60Hz.

Running MMS 2024 Suite v 23.0.1.182 (UDP3) and VPX 14 - v20.0.3.180 (UDP3)

M Audio Axiom AIR Mini MIDI keyboard Ver 5.10.0.3507

VXP 14, MMS 2024 Suite, Vegas Studio 16, Vegas Pro 18, Vegas Pro 21,Cubase 4. CS6, NX Studio, Mixcraft 9 Recording Studio. Mixcraft Pro 10 Studio. CS6 and DXO Photolab 8, OBS Studio.

Audio System 5 x matched bi-wired 150 watt Tannoy Reveal speakers plus one Tannoy 15" 250 watt sub with 5.1 class A amplifier. Tuned to room with Tannoy audio application.

Ram Acoustic Studio speakers amplified by NAD amplifier.

Rogers LS7 speakers run from Cambridge Audio P50 amplifier

Schrodinger's Backup. "The condition of any backup is unknown until a restore is attempted."

johnebaker wrote on 12/9/2018, 7:28 AM

@CubeAce

HI

. . . . Every 3dB you reduce, you are halving your volume . . . .

Not quite - a ±3dB change is a halving or doubling of the power level or sound intensity - this is not the same as perceived loudness.

For a perceived loudness doubling or halving the change is ±10dB. The perceived loudness of the sound depends on several factors: the amplitude, the sound pressure level, the frequency, and the time behaviour of the sound.

@discburn

I always set the sound levels to -6dB for DVD, BD and video files irrespective of their destination, however not using the meters in VPX.

I and several other users in the forum is use the MeldaProduction MDynamicEq to compress the audio. IMHO this also improves the clarity of the audio.

HTH

John EB

 

VPX 16, Movie Studio 2025, and earlier versions 2015 and 2016, Music Maker Premium 2024.

PC - running Windows 11 23H2 Professional on Intel i7-8700K 3.2 GHz, 16GB RAM, RTX 2060 6GB 192-bit GDDR6, 1 x 1Tb Sabrent NVME SSD (OS and programs), 2 x 4TB (Data) internal HDD + 1TB internal SSD (Work disc), + 6 ext backup HDDs.

Laptop - Lenovo Legion 5i Phantom - running Windows 11 24H2 on Intel Core i7-10750H, 16GB DDR4-SDRAM, 512GB SSD, 43.9 cm screen Full HD 1920 x 1080, Intel UHD 630 iGPU and NVIDIA GeForce RTX 2060 (6GB GDDR6)

Sony FDR-AX53e Video camera, DJI Osmo Action 3 and Sony HDR-AS30V Sports cams.

CubeAce wrote on 12/9/2018, 9:12 AM

@johnebaker

I was talking about the problems with clipping, which is about power levels and not perceived loudness.

From a broadcast perspective, years ago it depended on what the destination of the file was for, so for broadcast, a minus value of minus 4dBm using peak program meters was set or 0 if VU meters were in use, whereas directly to tape, depending on the machine, one could easily get into plus figures. The lowering of the signal in the recording was to protect the transmitter valves at the broadcast stage. I'm not quite sure what you mean by the time behaviour of the sound unless you are referring to phase variance or the problems with regards to sound absorption of frequencies or the problems with eigentones from reflective surfaces.

As far as I'm aware the -6dB is normal when dealing with 5.1 surround sound files.

 

Windows 10 Enterprise. Version 22H2 OS build 19045.5737

Direct X 12.1 latest hardware updates for Western Digital hard drives.

Asus ROG STRIX Z390-F Gaming motherboard Rev 1.xx with Supreme FX inboard audio using the S1220A code. Driver No 6.0.8960.1 Bios version 1401

Intel i9900K Coffee Lake 3.6 to 5.1GHz CPU with Intel UHD 630 Graphics .Driver version Graphics Driver 31.0.101.2135 for 7th-10th Gen Intel® with 64GB of 3200MHz Corsair DDR4 ram.

1000 watt EVGA modular power supply.

1 x 250GB Evo 970 NVMe: drive for C: drive backup 1 x 1TB Sabrent NVMe drive for Operating System / Programs only. 1X WD BLACK 1TB internal SATA 7,200rpm hard drives.1 for internal projects, 1 for Library clips/sounds/music/stills./backup of working projects. 1x500GB SSD current project only drive, 2x WD RED 2TB drives for latest footage storage. Total 31TB of 10 external WD drives for backup.

ASUS NVIDIA GeForce RTX 3060 12GB. nVidia Studio driver version 572.60 - 3584xCUDA cores Direct X 12.1. Memory interface 192bit Memory bandwidth 360.05GB/s 12GB of dedicated GDDR6 video memory, shared system memory 16307MB PCi Express x8 Gen3. Two Samsung 27" LED SA350 monitors with 5000000:1 contrast ratios at 60Hz.

Running MMS 2024 Suite v 23.0.1.182 (UDP3) and VPX 14 - v20.0.3.180 (UDP3)

M Audio Axiom AIR Mini MIDI keyboard Ver 5.10.0.3507

VXP 14, MMS 2024 Suite, Vegas Studio 16, Vegas Pro 18, Vegas Pro 21,Cubase 4. CS6, NX Studio, Mixcraft 9 Recording Studio. Mixcraft Pro 10 Studio. CS6 and DXO Photolab 8, OBS Studio.

Audio System 5 x matched bi-wired 150 watt Tannoy Reveal speakers plus one Tannoy 15" 250 watt sub with 5.1 class A amplifier. Tuned to room with Tannoy audio application.

Ram Acoustic Studio speakers amplified by NAD amplifier.

Rogers LS7 speakers run from Cambridge Audio P50 amplifier

Schrodinger's Backup. "The condition of any backup is unknown until a restore is attempted."

johnebaker wrote on 12/9/2018, 3:40 PM

@CubeAce

Hi

. . . . was talking about the problems with clipping, which is about power levels and not perceived loudness . . . .

This is not what you stated in the first sentence of the previous post as I quoted above:

. . . . . . . Every 3dB you reduce, you are halving your volume . . . . . . . .

Where volume and loudness are the same thing, hence my response.

It is very important to be specific when referring to volume/loudness, power levels and clipping.

John EB

VPX 16, Movie Studio 2025, and earlier versions 2015 and 2016, Music Maker Premium 2024.

PC - running Windows 11 23H2 Professional on Intel i7-8700K 3.2 GHz, 16GB RAM, RTX 2060 6GB 192-bit GDDR6, 1 x 1Tb Sabrent NVME SSD (OS and programs), 2 x 4TB (Data) internal HDD + 1TB internal SSD (Work disc), + 6 ext backup HDDs.

Laptop - Lenovo Legion 5i Phantom - running Windows 11 24H2 on Intel Core i7-10750H, 16GB DDR4-SDRAM, 512GB SSD, 43.9 cm screen Full HD 1920 x 1080, Intel UHD 630 iGPU and NVIDIA GeForce RTX 2060 (6GB GDDR6)

Sony FDR-AX53e Video camera, DJI Osmo Action 3 and Sony HDR-AS30V Sports cams.

discburn wrote on 12/10/2018, 1:36 AM

Thanks JohnEB and CubeAce,

Yes, it’s not actually a clipping problem, more a speaker 'rattling' or 'booming' problem, where no clipping at all is occurring. It does not at all occur when peaks are kept below -6dB. (And thanks for proper compression practice advice - duly noted.)

Interestingly, on a new LG TV, I checked a Netflix movie. It played fine when the TV volume was set at 30. I played one of my -6dB flies and had to reduce TV volume to 25 for same perceived loudness. So Hollywood keeps plenty of headroom, too!

It's YouTube that’s confusing it for me. The peaks are definitely set close to 0dB. And the smart TVs have to deal with that. I don’t want to have to make two versions of every file - one peaking at -.5dB for YouTube and one peaking at -6dB for 'TV viewing'.

What do you guys do when uploading to YouTube?

Seamus

RogerGunkel wrote on 12/10/2018, 4:13 AM

Are you sure that we are not confusing the wood for the trees here? I can produced a finished edit at what I consider to be the correct audio levels, let's say at peaks of a maximum of 0db VU. I then make a dvd and a usb of the finished work which I play back for testing on the P.C. However, when I play the dvd to the tv via my dvd player, I have to turn up the tv volume way above what I would to watch a direct tv channel. I then play back the usb directly in my tv and find that the volume I had set for the dvd is now way too high. I then watch a Youtube video and again need to reset the volume levels, then every channel seems to vary when I watch the tv.

The point I am making here is that whatever volume something is recorded at, does not mean that the different equipment you are playing it back on, or whatever playout gear the broadcasters are using, is going give you a standard volume.

During my years as an audio engineer in my studio, every piece of audio equipment was lined up with an audio tone and frequency analyser to set interfaced equipment at an optimum level. Sadly that discipline doesn't apply across the myriad interfaces between computers, tvs, playback equipment and transmission and playout equipment. There is also people's own perception of how they listen to audio. I can take a carefully balanced wedding video to a client, only to find that they have their tv plugged through a system that has been set up for their playstation with massive bass enhancement for war games, or weedy little speakers on a tiny tv that rattle at any reasonable volume.

The same applies to properly balanced colour, when put on someone's tv that is set up totally wrong. I feel like apologising for the over saturated colour, or grabbing the remote and resetting their tv.

Roger

discburn wrote on 12/10/2018, 8:20 AM

Hi Roger,

What you’re saying is very true. I believe different devices do indeed play out at different levels. A 'calibrated' programme coming from a 'calibrated' studio should indeed stand on its own.

Seamus