DVD authoring: choosing between MP2 and AC3 audio

[Elhombresombra]

Hi folks,

I need to create some video DVD from MP4 source files. The audio is NOT multilingual because it's just music, so there will be only one audio track, in stereo.

My goal is to keep the best possible quality of the final audio, because it's classical music.

In the MP4 source files, the audio was encoded (according to the "file properties" dialog) with the AAC codec.

The authoring program I'm using, DVDStyler, offers two options about the encoding of the final audio: MP2 or AC3.

My DVD players have the capability to correctly reproduce the MP4 files with AAC audio (just for information).

Given all the above, what option should I choose for the final audio, MP2 or AC3, and why?

Thank you!!!

[_Al_]

late for the party, some 15 years

limitations:
NTSC DVD -> go with AC3, but most players would have no problem with mp2 audio, especially today, but just guessing, I never tempted the fate, but as long you have a choice ...

AC3 is a bit more compressible,
if having more than 2channels definitely I would not complicate it and use AC3

[videobruger]

Just to mention: DVDStyler also offers PCM.


[Attachment 44537 - Click to enlarge]

[Elhombresombra]

late for the party, some 15 years

limitations:
NTSC DVD -> go with AC3, but most players would have no problem with mp2 audio, especially today, but just guessing, I never tempted the fate, but as long you have a choice ...

AC3 is a bit more compressible,
if having more than 2channels definitely I would not complicate it and use AC3

Al, I live in Europe, final DVD should play on PAL monitors... so?

[Elhombresombra]

Just to mention: DVDStyler also offers PCM.


[Attachment 44537 - Click to enlarge]

Sorry, I haven't reached to THAT menu so far... I just got stuck with the initial choice between MP2 and AC3.

I don't mind about adding more audio tracks (only one, music) nor file size (each programme is short, much shorter than a movie, so plenty of available space on a 4.7 GB DVD). I just want to avoid altering/tampering with the original audio as much as possible...

[ron spencer]

Always AC-3 and for 2 channel 192 rate is standard. PCM is ok for untouched too.

[_Al_]

so plenty of available space on a 4.7 GB DVD). I just want to avoid altering/tampering with the original audio as much as possible...

PCM means uncompressed, so whatever your original AAC was, you'll here the same, but bitrate is about 1500kbps. So grab some DVD bitrate calculator and maybe you have enough of room.
Those 1,5 Mbit is quite a lot.

If you encode to AC3 choose say 224kbs (but as was said 192 should be enough also) that is plenty for 2channels.
If you go with mp2 audio and say you encode to the same bitrate, in theory, it would be tiny bit worse. In theory. Not saying that you could hear any differences.

[pandy]

MP2 with sufficient bit rate can be marginally better than more common AC-3 - however you must have uncompressed source - for your case AC-3 is more compatible and overall will provide very good results so my vote is for AC-3 audio, 192kbps will be more than OK (if you insist you may go for 224 - 256kbps) but this will be IMHO waste of bits that can be used by video encoder to provide better quality (MPEG-2 usually works quite well above 5 - 6Mbps).

[Case]

If you go with mp2 audio and say you encode to the same bitrate, in theory, it would be tiny bit worse.

MP2 with sufficient bit rate can be marginally better than more common AC-3 - however you must have uncompressed source

Two seemingly contradicting opinions. Yesterday I tried to find comparison tests, but found nothing I felt comfortable quoting here. Perhaps it’s just too close with sufficient bitrate?

[lordsmurf]

I never compress to 192, trying to always use 224-384.

Either MP2 or AC3 is fine.

Authoring apps will nag you because MP2 not officially supported in NTSC, but I've never come across a player that reject MP2 audio.

[pandy]

If you go with mp2 audio and say you encode to the same bitrate, in theory, it would be tiny bit worse.

MP2 with sufficient bit rate can be marginally better than more common AC-3 - however you must have uncompressed source

Two seemingly contradicting opinions. Yesterday I tried to find comparison tests, but found nothing I felt comfortable quoting here. Perhaps it’s just too close with sufficient bitrate?

Being not native English i see nothing contradictory there - where Al wrote MP2 and AC-3 same bitrate then MP2 lower quality then i agree - AC-3 is more efficient and able to deliver same or similar quality with lower bitrate as such AC-3 with 192kbps will be better than MP2 at 192 however MP2 at 320kbps may be better than AC-3 at 320kbps (as AC-3 has own limitations related to more complex coding). MP2 may outperform AC-3 with sufficiently higher bitrate due different coding principle. That's why Musepack was founded on layer II technology not layer III (close to AC-3).
Once again - AC-3 is high quality and most compatible way of lossy audio compression for DVD.

[Elhombresombra]

Hello folks, I'm sorry for having to revive a thread that is almost 1 year old but... ok, "sh*t happens"!

While my source files were initially MP4 (see the opening post), now they have changed to MKV.
The reason is simple. The videos I need to burn as DVD video come in as two separate streams, video and audio. The video stream is MP4 (H.264 coec) while the audio is a WEBM file (OPUS codec). I have been explained that those two formats, due to compatiblity reasons, cannot be muxed to a MP4 final file. Instead, they CAN be muxed into an MKV container, hence I did that, and it worked!. :0)

So now the initial question is still the same ("Given all the above, what option should I choose for the final audio, MP2 or AC3, and why?") but the file to be burned is MKV instead of MP4... o_O

[KarMa]

I would be wary of any AC-3 encoder that is not the official Dolby AC-3 encoder, which comes with paid software. This is not because of my love of Dolby (quite the opposite), but just because the Dolby encoder is so much better when compared to the free Aften AC3 encoder, which I think is in DVDStyler and FFMPEG. Dolby is the one who created the AC-3 standard so they are going to naturally do better. The difference between Dolby and Aften was quite obvious to me in personal A/B tests. I never tested TwoLAME MP2 vs Aften AC3 but that would certainly be interesting, and I bet TwoLame would do the same or better than Aften.

Are these going to be data discs? As the DVD Video standard won't play MKV or MP4 files and will need to be re-encoded. Certainly won't accept Opus or H.264 either.

[Elhombresombra]

Are these going to be data discs? As the DVD Video standard won't play MKV or MP4 files and will need to be re-encoded. Certainly won't accept Opus or H.264 either.

I need to burn them as DVD-Video... they are intended to be watched on a standard domestic DVD player...

[davexnet]

If you're going to convert to mpeg-2 DVD the source format is irrelevant.
For NTSC DVD, mp2 may not be not fully compatible, I've seen this before on my own DVD player,
where an mp2 title was silent in my standalone player but worked fine on the computer

It's also mentioned in here
http://www.digitalfaq.com/guides/video/introduction-author-burn.htm

[Elhombresombra]

If you're going to convert to mpeg-2 DVD the source format is irrelevant.
For NTSC DVD, mp2 may not be not fully compatible, I've seen this before on my own DVD player,
where an mp2 title was silent in my standalone player but worked fine on the computerhttp://www.digitalfaq.com/guides/video/introduction-author-burn.htm

No no, all the DVDs will be PAL, not NTSC, so the issue you mentioned should not arise.
But the central question is still there: if the source file is now MKV (not MP4 like when I started the thread) and the two streams are H.264 and OPUS, what should I select when DVDStyler asks me to choose between MP2 and AC3??

[davexnet]

There is no technical reason you should pick one or the other, just your own personal preference.
Perhaps you should do A/B listening tests and see if you can hear the difference

[Elhombresombra]

There is no technical reason you should pick one or the other, just your own personal preference.
Perhaps you should do A/B listening tests and see if you can hear the difference

It's very difficult, if not impossible at all, to detect a difference all by ear. You should use a distorsion meter, or a lab-grade spectrum analyzer to measure distortion-generated harmonics from a pure sine wave test signal.

I was thinking more of some theoretical reason. For example, between MP2 and AC3, which of them does not involve resampling or any unnecessary transformation of the audio contents? The less you tamper with the source, the better sound you get...

[davexnet]

There is no technical reason you should pick one or the other, just your own personal preference.
Perhaps you should do A/B listening tests and see if you can hear the difference

It's very difficult, if not impossible at all, to detect a difference all by ear. You should use a distorsion meter, or a lab-grade spectrum analyzer to measure distortion-generated harmonics from a pure sine wave test signal.

I was thinking more of some theoretical reason. For example, between MP2 and AC3, which of them does not involve resampling or any unnecessary transformation of the audio contents? The less you tamper with the source, the better sound you get...

Is the content mostly music ? Is your source pristine quality? Why don't you just use PCM then there is no compression
As for theoretical reason's you can do your own research

[pandy]

There is no technical reason you should pick one or the other, just your own personal preference.
Perhaps you should do A/B listening tests and see if you can hear the difference

Doubt you can use lossy encoded source as reference... sorry - PCM as source is different topic, i can imagine Dolby native decoder higher quality but question is how many people can pay few thousands $ for Dolby encoder...
We must rely on those free implementations which are fair trade between quality and overall cost.
AC3 suffer from pre and post ringing where MP2 should be free from this (but it is less accurate in absolute frequency resolution which is obviously visible if pure sine tone are feed to MP2 encoders).

[Elhombresombra]

@Davenext, @pandy,

the content is 100% music, of the most "difficult" type (large pipe organs: dynamics going from whispering to literally earsplitting, frequency from the 16 Hz of pedal low C to almost ultrasonic...)
I have no control over the audio format because the recordings weren't made by me. The audio track comes in as a WEBM file, OPUS codec. I know it's not the optimum but, for that very reason, I feel it is imperative to avoid any unnecessary manipulation/degradation of the audio track, hence the necessity to choose the better option between MP2 and AC3.

[pandy]

@Davenext, @pandy,

the content is 100% music, of the most "difficult" type (large pipe organs: dynamics going from whispering to literally earsplitting, frequency from the 16 Hz of pedal low C to almost ultrasonic...)
I have no control over the audio format because the recordings weren't made by me. The audio track comes in as a WEBM file, OPUS codec. I know it's not the optimum but, for that very reason, I feel it is imperative to avoid any unnecessary manipulation/degradation of the audio track, hence the necessity to choose the better option between MP2 and AC3.

IMHO both (AC3 and MP2) can deliver similar quality (MP2 may be marginally better at bitrates around 256 kbps up and fast/transient rich music however this is valid only for lossless sources). For already lossy compressed audio i would not expect any difference and AC3 is slightly more compatible (market penetration) as Dolby marketing is very convincing and they charging Us for their decoder present everywhere...

[davexnet]

Can you provide a minimum 30 second sample of your source? Something that best
demonstrates the audio

[Cornucopia]

The problem is you're starting with an already "ruined" music recording, because whether they've been compressed to AAC or to Webm/Opus, they're compressed.
Now, if they were compressed to the highest possible compressed bitrate - say 320kbps - the loss you have already incurred might not be that much.
If the music quality is of the prime importance in this equation, I strongly recommend decoding and then leaving it as LPCM. At least then it will have only incurred the original loss.
I say this having had over a decade of professionally recording classical/sacred ORGAN performances, so I know what you would have been capable of at the start and what you now could get.
24/96 LPCM recordings played over a stereophile system is truly astounding.

Scott

[Elhombresombra]

@davexnet (sorry for the previous misspell of your username ) here you can find a typical WEBM file that I have to mux with the video. It's more than the 30 seconds you requested (actually it is a performance in whole) because it was easier to leave it "as is", the size is very small anyway. If you are looking for the loudest part, he begins to "throttle up" that thing seriously at about 6'15" into the file...

@cornucopia, when you say LPCM, do you mean the .wv (not .wav) format? It is a "lossless" choice that the conversion program offers me. Is that better than to leave the audio source in the original WEBM?...

[davexnet]

@davexnet (sorry for the previous misspell of your username ) here you can find a typical WEBM file that I have to mux with the video. It's more than the 30 seconds you requested (actually it is a performance in whole) because it was easier to leave it "as is", the size is very small anyway.

@cornucopia, when you say LPCM, do you mean the .wv (not .wav) format? It is a "lossless" choice that the conversion program offers me. Is that better than to leave the audio source in the original WEBM?...

Thanks for posting, great clip - I see what you mean about the dynamic range.
One thing to your advantage, it's already DVD compliant 48kHz sample rate and 16 bit depth.
Nothing to degrade the sound, just choose wav/pcm in your DVD creation program. The audio will be uncompressed
and used as-is. Your mp2 Vs. Ac3 was a false dichotomy.

Here4's how you set it up in AVStoDVD

[Elhombresombra]

Thank you very much!... Will try that ASAP! (now it's bedtime over here, not sure whether tomorrow it would be possible since it is Xmas eve... )