Opinions on converting audio to AC3 (besweet)?

[Lester Burnham]

I mainly use VideoReDo for captures from digital satellite, here in the UK, from a proprietary PVR (Sky+ box).

Sometimes it's movies, other times TV series that I want to keep.

In some cases, the movies are actually broadcast in DD, and the PVR box gets them as such and I can watch them using DD using my amp and / or TV (TV has DD and dts built in, and has an optical connection from the Sky+ box).

However, the capture device I use (Hauppauge PVR USB2) doesn't capture with AC3, only mpeg.

So what I was wondering is, whether it makes sense for films that are fairly long, to use VideoReDo to do the trimming, save to ES, then convert to AC3.

I've use besweet before for this occasionally (more for when small audio sections are to be used in authoring packages). Clearly I'd need to reduce the bitrate in order to see reduction in the audio size - and I accept the audio tends to be fairly insignificant in size when compared with the video.

But say I was to attempt to maximise the capacity of a DVDR, would it make much sense to consider this for films that have been captured and are a little bigger than the 4.7G single layer size?

And if anybody does do this sort of thing, do you have any issues with lip-synch, and what sort of AC3 bitrates do you use (in comparison to the mpeg ones).

I know I can shrink the final authored DVD using dvdshrink or Nero recode, and I know I can use custom bitrates when I capture, to try and get it to fit - but I have capture settings that I'm happy with, and tend to do fairly big captures each time (5,6,7 or 8 hours worth).

[DanR]

We will be offering MPEG audio bit rate re-encoding in Q2. We might also be offering AC3 output as an option. Depends on whether we want to pay for the encoding licenses.

[Lester Burnham]

We will be offering MPEG audio bit rate re-encoding in Q2. We might also be offering AC3 output as an option. Depends on whether we want to pay for the encoding licenses.



That would be a nice touch to incorporate it - it would certainly address any lip-synch issues, which VideoReDo has been something of a panacea for.

AC3 would seem a good idea - mainly because it's meant to offer good results from reduction in bitrate - to do it mainly within mpeg, doesn't seem as much of a win as I could capture with lower audio bitrate - but to convert and not lose as much, and do it within the one product / step would be great.

[DanR]

Do you know how much AC3 reduces filesize for equivalent audio quality when compared to MPEG Layer 2 audio?

[bitter_old_man]

They are about the same size. I took a 47.2 MB wav and converted it to mp2 and ac3 (usingBeSweet). Both were encoded at 256 kbps. The mp2 file was 256 bytes larger.

Barry

[phd]

I believe there have been a few discussions and comparisons about it on the Ulead forum. File size was roughly the same.

The major advantage was that AC3 was compliant for NTSC DVD players. PAL standard includes MPEG audio whereas the NTSC standard does not.

I could probably find the threads again if desired.
http://phpbb.ulead.com.tw/EN/viewtopic.php?t=763

[Lester Burnham]

I believe there have been a few discussions and comparisons about it on the Ulead forum. File size was roughly the same.

The major advantage was that AC3 was compliant for NTSC DVD players. PAL standard includes MPEG audio whereas the NTSC standard does not.

I could probably find the threads again if desired.
http://phpbb.ulead.com.tw/EN/viewtopic.php?t=763

As something of a follow up to all the replies on file size - I think the point being that for the same bitrate, you're not going to see much difference by converting domestically.

However, the theory being that you should be able to get away with a reduction in bitrate, but not feel the pain as much in sound quality.

[BogieV]

[quote=phd]
However, the theory being that you should be able to get away with a reduction in bitrate, but not feel the pain as much in sound quality.

Yeah, I've read that AC3 at 192kbs is pretty good compared to 224kbps MP2.
But how about the pain going through the conversion process?

I'm in Japan and here is very NTSC country but all the TV capture cards are MP2 audio only (no AC3 as in the standard for NTSC).
Haven't heard anybody complaining his DVD player is not capable of playing it.

I also tend to use 384 or 320 kbps as I don't like the big drop in quality compared to the original sound while slight lowering of the video bitrate doesn't change the PQ dramatically.

[Lester Burnham]

[quote=phd]
However, the theory being that you should be able to get away with a reduction in bitrate, but not feel the pain as much in sound quality.

Yeah, I've read that AC3 at 192kbs is pretty good compared to 224kbps MP2.
But how about the pain going through the conversion process?

I'm in Japan and here is very NTSC country but all the TV capture cards are MP2 audio only (no AC3 as in the standard for NTSC).
Haven't heard anybody complaining his DVD player is not capable of playing it.

I also tend to use 384 or 320 kbps as I don't like the big drop in quality compared to the original sound while slight lowering of the video bitrate doesn't change the PQ dramatically.

Well I capture in MPEG (hardware MPEG encoder - doesn't do AC3, though). And I always use 384 for audio.

If I convert (by first demuxing to ES) the audio to AC3 using besweet, if I use like-for-like bitrates, I'll just see (roughly) the same size file.

I guess what I was wondering is, for captures (say films) that are pushing the limits (or perhaps exceeding them) of a single layer, whether people have played with conversion to AC3 with a reduced bitrate, to squeeze a bit more on the disk (perhaps in order to not have to reduce the video bitrate as much), and if so, whether they've noticed much difference in the audio quality - or other things like lip synch.

[BogieV]

Yes, exactly! You can get as low as 192kbps with AC3 without noticing quality drop - that's the whole idea.
Don't know about lip synch and haven't tried it by myself though.
Try it and let us know how you feel about it!

[Anonymous]

I normaly always author with a re-encoded ac3 stream as the main audio track.
And include the original demuxed audio stream as a second audio track. So that i can always make the file back into an original mpg again if needed, much like a backup for the original mpg file.

The ac3 i normaly re-encode to 448kbps from the original mpg file audio stream, as can be done from TMPGENC ac3. This i do from any bitrate even 192kbps. Have thought many times if re-encoding the original mpg file audio stream first high as possible and then convert to ac3. Maybe it would make the ac3 audio better again. any ideas ?

I make the assumption the bigger bitrate, better the final audio quality.

[Lester Burnham]

I normaly always author with a re-encoded ac3 stream as the main audio track.
And include the original demuxed audio stream as a second audio track. So that i can always make the file back into an original mpg again if needed, much like a backup for the original mpg file.

The ac3 i normaly re-encode to 448kbps from the original mpg file audio stream, as can be done from TMPGENC ac3. This i do from any bitrate even 192kbps. Have thought many times if re-encoding the original mpg file audio stream first high as possible and then convert to ac3. Maybe it would make the ac3 audio better again. any ideas ?

I make the assumption the bigger bitrate, better the final audio quality.

You say you normally re-encode to 448 for AC3 - what do you start with?

There's nothing gained - quality wise by re-encoding with higher bitrates - they need to be high at source. I can understand retaining a high bitrate track, and dropping the rate where it suits, though.

[DanR]

What has been ignored in this discussion is the optimum bit rate to encode the source material. Here in the US, most OTA (over the air) signals have a FM quality sound. This seems to imply that an MPEG Layer 2 bit rate of 192 to 228 should be more then sufficient to accurately capture the entire audio stream. Bit rates above that are wasted.

[Lester Burnham]

What has been ignored in this discussion is the optimum bit rate to encode the source material. Here in the US, most OTA (over the air) signals have a FM quality sound. This seems to imply that an MPEG Layer 2 bit rate of 192 to 228 should be more then sufficient to accurately capture the entire audio stream. Bit rates above that are wasted.

To be fair, though, does depend on source.

Here in the UK, there's terrestrial analogue, terrestrial digital, satellite digital and cable digital.

Some sources could be pure analogue, NICAM stereo. Other digital broadcasts can vary in bitrate, and on digital satellite can be DD / AC3.

I get your point though - it's as daft to capture with a higher bitrate than is represented in the carrier, as to re-encode with a higher bitrate after it's been encoded.

[Anonymous]

@Lester and readers

I capture via dtt device at 192 or 256 i leave it as such on the mpg. But when demuxing for DVD i also include an ac3 track that i encode at 448kbps would go higher if i could like 600+kbps that besweet can do. But can't as DVDLab doesn't accept them and is limited to 448kbps limited by tmpegenc ac3.

For music videos that i have not yet started to put on DVD. They were recorded from dtt at 192 or 128 and used another editor to make them to 384kbps. But will also make these ac3 at 448kbps when authoring with DVDLab.

Bigger must be better, well i'm hoping this is so, with what i do above.

[DanR]

Bigger must be betterNot true at all. You may just be wasting space.

[Anonymous]

Is that it never will make the audio quality better or will it just sometimes make the audio quality better ?

[DanR]

Is that it never will make the audio quality betterYou can never improve on the original. You can however, make it worse by re-encoding to too low a bit rate.

[Anonymous]

That must mean the sound of the original Mpeg1 layer 2 is locked. Which i believe is not the same with other formats such as wave and mp3 for audio. Glad i have never reduced any bitrates in mpgs so far.

Dan why is it that other formats can be improved upon whereas our mpgs cannot be audio improved ?

[Lester Burnham]

@Lester and readers

I capture via dtt device at 192 or 256 i leave it as such on the mpg. But when demuxing for DVD i also include an ac3 track that i encode at 448kbps would go higher if i could like 600+kbps that besweet can do. But can't as DVDLab doesn't accept them and is limited to 448kbps limited by tmpegenc ac3.

For music videos that i have not yet started to put on DVD. They were recorded from dtt at 192 or 128 and used another editor to make them to 384kbps. But will also make these ac3 at 448kbps when authoring with DVDLab.

Bigger must be better, well i'm hoping this is so, with what i do above.

There is absolutely no point in re-encoding with a higher bitrate. None, nada, zip, bubcas.

You can't encode that extra "quality" back in, you have to capture / extract at the best rate you want.

And to add to this - Dan raises a fair point - there's no point capturing / encoding at a greater bitrate than the transport is supplied with.

[Lester Burnham]

That must mean the sound of the original Mpeg1 layer 2 is locked. Which i believe is not the same with other formats such as wave and mp3 for audio. Glad i have never reduced any bitrates in mpgs so far.

Dan why is it that other formats can be improved upon whereas our mpgs cannot be audio improved ?

Because it's a lossy compression.

[DanR]

Dan why is it that other formats can be improved upon whereas our mpgs cannot be audio improved ?

All lossy compression techniques like MPEG audio and video, AC3 audio, AAC audio, etc. discard information to achieve their compression objectives. The information they discard is based upon "models" developed from knowledge of how people see and hear. The lower the bit rate the more information is lost.

WAV / LPCM are not compressed (technically they are because all digitized material is compressed due to sampling, but that's another discussion [CD vs. LPs]). Therefore there is more quality to be preserved by using a high bit rate in the compression.

In my own case, I have moved much of my CD collection to a media server. I use the lame MP3 encoder with 384 Kbps variable compression. The 384Kbps does an excellent job (not perfect) and the variable compression aspect reduces the bit rate if the whole 384 Kbps isn't needed. Neither AC3 nor MPEG Layer 2 offer variable bit rates (thank god).

[Anonymous]

I have never yet come accross an ideal way to preserve audio CD quality in any pc compressed format. But maybe just maybe sometime it might happen, then i'll transfer them all to DVD. So reducing the space taken up by of many CDs.

[DanR]

There is no ideal way to preserve CDs with lossy compression, its always a trade-off. However, there's an lossless compressor, I think its called "Monkey Audio" that achieves a 50% compression with no quality loss at all.

You might want to check out the Hydrogen Audio message board. Its a great board for audio compression discussions.

[bitter_old_man]

In addition to Monkey's audio, there are a couple of other lossless codecs: shorten (SHN) and FLAC (Free Lossless Audio Codec). I back up rare CDs by ripping the CD to a cue file + wav (with EAC) and then compress the wav, usually with FLAC.

Barry

[Anonymous]

Thanks for the tips all i will look into them for more info.

Ever heard of a way to convert a two channel file into a 5.1 surround audio wave file. Had a look at besweet but am not to sure. Such as convert into six files and then each adjust the volume. Then stick them back together to make a 5.1 surround audio track. I remember trying to do it once but it didn't want to do it from a 2 channel wav file as i remember, any ideas ?

[bitter_old_man]

Here's a guide for converting stereo to 5.1: http://forum.doom9.org/showthread.php?s=124aa48840e7ee8a42755b77130abbaf&threadid=85446

It requires an audio editor that supports VST plugins. Be sure to read the entire thread, as there are some suggestions on doing it with free software.

Does it work? I don't know, I've never tried it.

Barry

[Anonymous]

Worth a try, i'll go and read that now thanks :roll:

[Anonymous]

:shock: this might be good for converting 2 channel mpg files in to 5.1 for DVD authoring as well as what i need for CD to 5.1.

V.I is a 2-input, 6-output VST plugin that converts a two-channel stereo input to a six channel output conforming to the ITU standard for 5.1 of Left, Right, Center, LFE, Surround left, Surround right.

Will go and read the rest, i'm thinking Audacity uses VST plugins, so this might be a breeze to do

Nice find Barry ;)