Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: [TOS #5 + DNDtR again] From: IETF Opus codec now ready for testing (Read 2463 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

[TOS #5 + DNDtR again] From: IETF Opus codec now ready for testing

Just a fast question, not really fitted here, but not worth a new topic i think.

When encoding Mono and Stereo, is Stereo Twice of Mono as in "reality"?

With that i mean

128kbps for Mono is like 256kbps?

And with that i also wonder, what Uncoupled is in Opus when it´s tells me the number of channels.

[TOS #5 + DNDtR again] From: IETF Opus codec now ready for testing

Reply #1 –
Stereo need twice the bitrate of mono? Most probably not.

For efficiency, modern codecs at least try to convert the left and right channel to their sum (mid) and their difference (side). If you have a sound source from the front (very probable for speech, but often also for music which does not use a lot of phasing and surround effects), then both left and right channel will have very similar content, so their sum will have "a magnitude" more impact and needs more precision than their difference.

MP3 already has "Joint Stereo" as a kind of channel coupling. I guess there can be more elaborate decisions, e.g. per frequency subband, if left/right or mid/side encoding is more efficient for a specific audio frame.

I once read about Dolby Digital (AC3) in a book about DVDs that their channel coupling is efficient enough to require only the square root of the channel number as factor of the mono bitrate for an acceptable quality; that would mean a comparable quality between 160 kbps mono, 224 kbps stereo (160*√2), and 384 kbps discrete 5-channel + LFE (160*√5+lowband). A half-baked scheme? ... The audience may decide.

[TOS #5 + DNDtR again] From: IETF Opus codec now ready for testing

Reply #2 –
Quote from: zerowalker on
Just a fast question, not really fitted here, but not worth a new topic i think.
Wrong. As I already said last time, starting a new thread for such a question, even if you expect the answer to be simple, is much better than clogging a thread up with things that aren't on topic.

Also, starting a new thread would make your question visible to the rest of the people at HA who aren't following this thread, and give people who are interested in answering such questions a chance to respond. The only people who are following this thread are those who are interested in the testing and development of Opus, and they're here to talk about the testing and development process, not to parrot basic reference materials to those who are unwilling to look up answers to their own questions.

[TOS #5 + DNDtR again] From: IETF Opus codec now ready for testing

Reply #3 –
Okay so in a Easy way of speaking, it will use the same bits for both channels if the sound is the same, and the differences on their own?

Sound very neat, for normal musics etc, it probably is just up to 50% more bits needed for Stereo than Mono i guess:)


But how is it going for Opus, when is it planned to be Final?
as it´s RC currently, don´t mean to complain, i just wonder, i think i read it but can´t find it:S

[TOS #5 + DNDtR again] From: IETF Opus codec now ready for testing

Reply #4 –
It’s only because you closed with a question that’s actually relevant to Opus that I am not binning the three posts about joint stereo. For the latter, there are shedloads of information here and elsewhere that can answer any questions you might have, especially if they are as basic as these. Please stay on-topic now; make another thread if you must, but only if you’ve made some attempt to find things out for yourself beforehand.

[TOS #5 + DNDtR again] From: IETF Opus codec now ready for testing

Reply #5 –
Quote from: zerowalker on
When encoding Mono and Stereo, is Stereo Twice of Mono as in "reality"?

With that i mean

128kbps for Mono is like 256kbps?


I have given this some thought myself.

It is worth remembering that, when first introduced, Stereo was thought of as a gimmick by some, similar to the way 3D Television is at the moment.

Phil Spector popularised the phrase "Back To Mono" and John Lennon could be seen wearing a "Back To Mono" badge in the late 60s.

Spectors classics were recorded in mono and I think Spector did not like the separation that Stereo gave you and preferred a muddier feel where all of the sound merged together as if it were one big instrument.  Obviously, given that Spector is, by far, the most celebrated record producer in history, Phil's preference for mono deserves some consideration.

I know often 1950s acts had their classic recordings later reissued in "reprocessed stereo" and that this was really badly received by fans and by the music press.  I remember Rolling Stone in the 1970s saying to avoid buying a Chuck Berry LP at all costs because, although every performance was a classic, they were in reprocessed stereo.  To get the original mono version became the holy grail.

I've noticed when using the --downmix option with oggenc to downmix to mono vorbis that you can save a lot of space.
It also takes approximately half the time to encode to mono vorbis as it does to encode to stereo vorbis.

I did a test using Aretha Franklin's I Never Loved A Man, because the CD contains both the original mono version and a stereo version as a bonus track.
Note that on CD the mono mix is actually 2 identical stereo channels, as all CDs have to be in stereo I believe.
Test used Xiph's libvorbis encoder.  I used quality level 8 as this gives a target bitrate of 256kbps in stereo, which is the number you used in your question.
Quality level 8 is considered overkill by most people in most situations as many people on these forums find q4 transparent (especially when using the aotuv encoder), as the following link demonstrates:

http://www.hydrogenaudio.org/forums/index....showtopic=51289

MONO MIX TO MONO VORBIS

oggenc --downmix -q 8 03-i-never-loved-a-man-the-way-i-love-you-.flac
Opening with flac module: FLAC file reader
Downmixing stereo to mono
Encoding "03-i-never-loved-a-man-the-way-i-love-you-.flac" to
        "03-i-never-loved-a-man-the-way-i-love-you-.ogg"
at quality 8.00
   [ 99.5%] [ 0m00s remaining] -

Done encoding file "03-i-never-loved-a-man-the-way-i-love-you-.ogg"

   File length:  2m 51.0s
   Elapsed time: 0m 03.0s
   Rate:        57.4216
   Average bitrate: 123.5 kb/s

MONO MIX TO STEREO VORBIS

oggenc -q 8 03-i-never-loved-a-man-the-way-i-love-you-.flac
Opening with flac module: FLAC file reader
Encoding "03-i-never-loved-a-man-the-way-i-love-you-.flac" to
        "03-i-never-loved-a-man-the-way-i-love-you-.ogg"
at quality 8.00
   [ 99.5%] [ 0m00s remaining] -

Done encoding file "03-i-never-loved-a-man-the-way-i-love-you-.ogg"

   File length:  2m 51.0s
   Elapsed time: 0m 05.4s
   Rate:        31.6371
   Average bitrate: 190.5 kb/s

STEREO MIX TO MONO VORBIS

oggenc --downmix -q 8 13-i-never-loved-a-man-the-way-i-love-you-stereo-version-.flac
Opening with flac module: FLAC file reader
Downmixing stereo to mono
Encoding "13-i-never-loved-a-man-the-way-i-love-you-stereo-version-.flac" to
        "13-i-never-loved-a-man-the-way-i-love-you-stereo-version-.ogg"
at quality 8.00
   [ 99.6%] [ 0m00s remaining] -

Done encoding file "13-i-never-loved-a-man-the-way-i-love-you-stereo-version-.ogg"

   File length:  2m 46.0s
   Elapsed time: 0m 03.0s
   Rate:        56.0601
   Average bitrate: 123.5 kb/s

STEREO MIX TO STEREO VORBIS

oggenc -q 8 13-i-never-loved-a-man-the-way-i-love-you-stereo-version-.flac
Opening with flac module: FLAC file reader
Encoding "13-i-never-loved-a-man-the-way-i-love-you-stereo-version-.flac" to
        "13-i-never-loved-a-man-the-way-i-love-you-stereo-version-.ogg"
at quality 8.00
   [ 99.8%] [ 0m00s remaining] -

Done encoding file "13-i-never-loved-a-man-the-way-i-love-you-stereo-version-.ogg"

   File length:  2m 46.0s
   Elapsed time: 0m 05.5s
   Rate:        30.4371
   Average bitrate: 234.3 kb/s

So using the --downmix switch can almost half the average bitrate that is used (from 234.3kbps to 123.5kbps).
If the original source material is mono, however, using the --downmix switch does not save as much space as the encoder is intelligent and so, even when encoding into stereo, the average bitrate is considerably less (190.5kbps) than when encoding from a true stereo mix (234.3kbps).
Also, it appears that you cannot lower bitrate using the downmix options to such a degree at lower quality levels as you can at higher ones, presumably because, at lower quality levels, already everything is being done to try to minimise bitrate and, for instance, lossless coupling is not used by default at lower quality levels, as it would be considered too wasteful.

target bitrates (source: 'ogginfo' command)

q2 mono 70kbps  stereo 96kbps
q3 mono 80kbps  stereo 112kbps
q4 mono 86kbps  stereo 128kbps
q5 mono 96kbps  stereo 160kbps
q6 mono 110kbps stereo 192kbps
q7 mono 130 kbps stereo 224kbps
q8 mono 140kbps  stereo 256kbps

q2 stereo has the same target bitrate (96kbps) as q5 mono so if you are pressed for space on a portable device but do not want to encode to a lower quality level it seems like downmixing to mono could be an option.

Obviously, nearly all music before the 1960s was in mono, so, if you own a lot of older music, the --downmix switch could be handy there too.

I think it is worth mentioning these things as a lot of people are not aware of the --downmix switch but it can be handy.

Now opus is a bit different to vorbis, as you do not specify a quality level with opus, but rather a target bitrate.

I am not sure what average bitrate mono in opus would equate to the same quality at a given bitrate in stereo.

Does opus use lossless coupling at higher bitrates like vorbis does?  If so, what is the threshold at which lossless coupling becomes activated?  Would be interesting to know.
 

[TOS #5 + DNDtR again] From: IETF Opus codec now ready for testing

Reply #6 –
Quote from: db1989 on
It’s only because you closed with a question that’s actually relevant to Opus that I am not binning the three posts about joint stereo.

I take that back.

Congratulations, mamboman.