AC3 to mp3 using Azid & Lame : what are the best parameters with AC3 mono tracks ?

[Simplet]

I'm currently looking for the best way to decode an AC3 mono track to a mono mp3 file using Azid (AC3-> Wav) then Lame (Wav -> Mp3).

Because the AC3 file is only mono, I assume that the -L -3db switch in Azid is'nt useful, because it is intended to downmix 5.1 audio files.

My aim is to encode a mp3 file as small as possible, (i.e. in mono) and I'm rather surprised by the fact that even if my mono AC3 file is encoded in mono, the previous wave files I got were much bigger (more than 900 MB) than the original AC3 (150 MB). Therefore, even if I encode an mp3 in mono, the result is still a pretty huge file (around 80-100 MB).
For example, using the --alt-preset 128 switch for lame, which uses a joint-stereo encoding, gives me a 85 MB mp3, while --alt-preset 128 -m m, meant to encode in mono, gives me a 100 MB mp3 file (strange thing I can't explain).

So here are my questions :

What are the best Azid paramaters to decode an AC3 192 kbps mono file to a mono wav file ? Will it be really smaller ?
Is really mono encoding worth the pain using Lame because it gives me bigger files than joint stereo encoding for the time being ? If it is, what's wrong with the switches I used ?


I've already seen topics dealing with that matter on the forum but I didn't find an answer that could help me, so please, if you know how to do, it would be kind of you to help me ...

TIA

[DJ Bobo]

1) Decode using the following command line:
azid -c normal audio.ac3 audio.wav

2) Open your WAV file in WaveLab or CoolEdit, convert it to Mono (if it is showed as stereo), THEN AND ONLY THEN normalize it to 0 db (that's 100%). Then sample it down to 44,1KHz or 32KHz (32KHz if it is an old soundtrack, example: old anime like Dragon Ball Z)

3) Encode it in RazorLame to 64kbps or 80kbps Mono, that's it.

[Simplet]

Could you explain me why I should downsample the wav file to 44,1 khz or 32 khz ?

Thanks for your piece of advice.

[DJ Bobo]

For better sound quality!

But don't downsample to 32KHz if your soundtrack has frequencies over 16KHz!

You should ALWAYS downsample to 44,1KHz, this will make your mp3 files sound better by the same bitrate.

As said downsampling to 32KHz only if your movie is old and the sound definitely lacks on higher frequencies

[MaTTeR]

Originally posted by bobotns
You should ALWAYS downsample to 44,1KHz, this will make your mp3 files sound better by the same bitrate.

Are you talking about downsampling in general or just for mono targets? I don't really have experience with mono encodes but in general the only real reason to downsample is for much older cards that can't support 48kHz IMHO. Someone please correct me if I'm out of line...well except for ChristianHJW:D

EDIT- Downsampling to 32kHz will quite possibly cause sync issues during playback.

[DJ Bobo]

@ MaTTer
I recommend downsampling to 44,1KHz generally, because the mp3 will sound better.
You can test it yourself. Make an mp3 @128kbps, one time with 48KHz and one time with 44,1KHz. You will notice that the 44,1KHz version is cleaner and has less artifacts.
I noticed the biggest difference between the 2 versions with the Dolby Digital Canyon trailer. The 48KHz-mp3 version of it was like shit. The 44,1KHz-mp3 version of it was clean.

And downsampling to 44,1KHz doesn't affect the raw sound quality, since AC3-streams have only frequencies up to 20,30KHz. That means a cut @ 22050Hz won't deterior the sound.

Those who say don't downsample to keep the audio quality are kidding their selves! there is NO QUALITY LOSS when downsampling to 44,1KHz! and that IMPROVES the sound of the resulting mp3.

About 32KHz stuff, I've made many clips with 32KHz, there is NO SYNCH ISSUES!

NB: I always use WaveLab to downsample.

[MaTTeR]

Originally posted by bobotns
[B]@ MaTTer
I recommend downsampling to 44,1KHz generally, because the mp3 will sound better...You will notice that the 44,1KHz version is cleaner and has less artifacts.
I really don't understand why the 48kHz gives you artifacts. The Canyon Trailer is crystal clear on my home theater system @ 48khZ and 44kHz. What sound card are you using? Does it not support proper 48kHz? Latest drivers installed, etc? I definitely disagree with a 44kHz sounding better than a 48kHz. Although I sometimes can't hear the difference between 44 & 48kHz, I see no reason to downsample for quality reasons as you suggest. Why not save yourself some time and skip the step?


About 32KHz stuff, I've made many clips with 32KHz, there is NO SYNCH ISSUES! That's a bold statement considering several people have reported sync issues in the past;)

How does WaveLab downsampling look compared to SSRC?

[DJ Bobo]

My drivers are all ok...

Well, if you're not noticing this @128kbps, then encode to 96kbps. One time with 48KHz and one time with 44,1KHz.
If you don't find that the 44,1KHz-Version sounds better than the 48KHz, you have definitely hearing problems!

We can also speak theory if you want: notice that compression of a 48KHz-track is bigger than the compression of a 44KHz-track.
For example @ 128kbps: it's 11:1 compression if the WAV has 44,1KHz, but it's 12:1 in case of 48KHz-WAV ;)

About 32KHz: I told you, there is NO SYNCH ISSUES!
There is also many people claiming, that the sound is out of synch (even when it's 44 or 48KHz), I never had such problems, NEVER!
I don't even understand how people manage to get sound out of synch, I really don't understand this!

One can become synch issues if the mp3 ist not MPEG1 but MPEG 2.5 for example.
But claiming that 32KHz will result in synch issues is a great mistake. I tested my CDs with 32KHz sound on more than 30 computers, there was NO ISSUES in any of them.

I never used SSRC, since it's rumored to be quite slow. If it's faster than WaveLab, just let me know, I will change to SSRC immediatly!

And comparing 48KHz-WAV and downsampeld 44,1KHz-WAV: I never heard a difference!

[MaTTeR]

@bobotns
Like I said, I've never had a need to downsample to 32kHz so I'll take your word that it doesn't cause sync issues. Other users must be doing something wrong(Video stretch?).

If downsampling to 44kHz then of course the file will be smaller. Not enough difference to really even mention though. On a 10min Ogg Vorbis file, the difference between the 2 sampling rates is 70kb. I suppose this might be the only reason for users to downsample, just to save 2-3MB of audio space on a 1CD rip. It's really not going to save you very many bits. That's fine, where I take issue is that your saying the 44kHz will always sound better. This simply isn't true. Just because your system displays artifacts at 48kHz doesn't mean everyone else will get them. So maybe you could have worded it...I always use 44.1kHz because blah blah blah?

Again, I ask you what sound card model your using? Are you using a SoundBlaster not capable of handling 48kHz frequencies? In regards to SSRC, I have no idea what you would consider slow compared to WaveLab. If you notice, BeSweet and HeadAC3he use SSRC exclusively if that's any indication how well it works;)

[DJ Bobo]

@ MaTTeR
You really should make the 96kbps-MP3 test, it can't fail, the difference is absolutely clear, the 44KHz-track will sound better, definitely!

BTW, I have a YAMAHA YMF740C sound card

And as said it's not because of my drivers or something, 48KHz MP3s with higher bitrates (160kbps and up) sound normally. Some 128kbps do sound similiar in 44 and 48KHz, but in specific cases I can hear the better quality of the 44KHz-track.
48KHz-WAVs and DVD-Sound are also flawless.
So there is no way that my sound card is bad or something, it can handle 48KHz flawlessly.

About SSRC: it's rumored to have the best downsampling capabilities. That's why I'll not come here and say that WaveLab is better in that matter. But as said, WaveLab makes an excellent job in record time (Speed: about x15 for a stereo-WAV on my PIII @667MHz) and I can't hear ANY difference between 48KHz-WAV and 44,1KHz-WAV.

And I don't use BeSweet and such stuff, I do all things all by my self, step by step, because I simply don't trust any automation and/or all-in-one programs (it's so, I can't do anything about it!)

[MaTTeR]

LOL. How about I just take your word for it that the 44kHz sounds better at 96kbps. I simply don't have the time ATM. Maybe I can get to it in a couple of weeks(hopefully).

Very nice to hear about the Yamaha card. I've heard nothing but good things about them:)

This is somewhat of a mute discussion on my behalf as I no longer use MP3 anyways. Vorbis is the future my friend;) Smaller file sizes and better quality than the equivalent bitrate in MP3.

[DJ Bobo]

I tested that Vorbis format. It's quite good, I got nice results already with 96kbps.
But as you said, it's the future, the format is too new and needs to reach mature before I consider using it "officially".
I'm also waiting for a good tool that can mux ogg files with avi, without taking the hassle of Graphedit.
I'm not saying that Graphedit is bad, I'm just saying that it's not as simple as opening NanDub, select the files and click save ;)

[MaTTeR]

Originally posted by bobotns
I'm not saying that Graphedit is bad, I'm just saying that it's not as simple as opening NanDub

I can't agree with you more:) GE is not the most intuitive tool around but for the moment it works and is fast.

BTW- Compare that 96kbps Vorbis with the MP3 equivalent in a listening test. Yummy...Vorbis;)

[DJ Bobo]

That was just what I was saying ;)