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: how much does stereo "cost" ? (Read 16572 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

how much does stereo "cost" ?

Hi folks.

I apologize in advance, but I am a little bit confused about this. So instead of picking on this possibly short-sighted question I would hope you could help rephrase it or guide my unclear thoughts in the right direction.

The question is: how much does stereo "cost" in terms of file size ?

I will try to explain what I mean by this. I have a large collection of talk radio broadcasts (streamed or recorded on air) which I want to store in a lossy format on my HDD (I have not yet decided what codec). Since the content is not music but spoken words, I don't target transparent quality but will rather choose a lower bitrate, giving me smaller file sizes. In this situation, I asked myself whether or not to opt for stereo. Sure, stereo sound is more gratifying but I have to weigh this benefit against the additional costs in file size. Therefore I want to know how much we are talking about.

Obviously a stereo file will be bigger than its mono mix but only up to twice as big in the worst case scenario, so the ratio would be somewhere between 1.00 and 2.00.

[blockquote][blockquote][blockquote][blockquote](Although - if I'm not mistaken - the following post suggests that factors > 2.00 are possible, which to me seems against all logic ...

Quote from: Northpack on
... the main reason for me to encode supposed mono records to mono MP3s is that many (about the half really) mono records on CD have been expanded with some fake stereo effects. These effects very effectively knock out the channel correlation handling in joint stereo mode. Thus you will get a better (=more original) sounding file at ~90kbps (comb-filtered fake stereo reversed by downmixing to mono) than the ~200kbps file you'll get by just encoding it as joint stereo (I really don't know why, but those comb filter files will bloat even more than by factor 2).
)[/blockquote][/blockquote][/blockquote][/blockquote]

Results will certainly depend on the chosen codec and settings, so I fear a general answer to this question is not possible.

But even if we assume we have chosen a codec, how can the additional costs be measured ?
A) If a constant bitrate is used, say 128 kbps CBR and joint stereo, I can only *guess* what bitrate is required for the mono file to have the same quality.
B) If a variable bitrate is used, say 128 bkps VBR and joint stereo, I can select a quality level, then use the same quality level for the mono encoding and compare file sizes. This approach has been used here: http://www.hydrogenaudio.org/forums/index....st&p=665611 .But this begs the question if e.g. quality 6 for a stereo file is the same as quality 6 for a mono file. After all "quality levels" aren't an exact science. And how would one determine if a stereo file has the same quality as a mono file anyway since they sound differently? [blockquote](Example: a 7 seconds mp3 joint stereo sample in -VBR -V 6 results in 106kb and the same sample in mono, still with -VBR -V 6, results in 64kb ... but can we really say that stereo "costs" 42kb in this case? If I use a music test sample it is impossible for me to compare the quality of a mono encoding against a stereo encoding. And if I use a spoken word test sample, I can only ABX high bitrate differences. So given that my ears don't help me in finding a pair of 1 mono and 1 stereo file with the same quality to see the file size difference, isn't there any mathematical approach which could give me an answer somehow?)[/blockquote]

And what about codecs which don't have anything like joint or m/s stereo, but only dual channels? In this case the ratio would be close to 2.00 right ?

how much does stereo "cost" ?

Reply #1 –
As a side-question: if the talkradio broadcasts I mentioned basically consist of one guy speaking and sitting in front of a stereo mic, what difference is there actually between a mono and a stereo recording? My guess is that if the person moves his head from left of the mic to the right, you would notice this movement reflected in the levels of the L and R speakers. But if we assume the person doesn't change his position at all and is centered in front of the mic, there should not be any noticeable difference between a mono and stereo recording, right? And if the speaker sits somehow on the left of the mic, both speakers would basically emit the same signal, but at a lower level on the R speaker and that's all to the story ....

.... OR is there supposed to be a spatial effect by which I mean (for a lack of a better description) you can somehow hear/feel like you were in the radio studio or hear/feel like the radio host were in your room while losing this effect completely in mono ?

how much does stereo "cost" ?

Reply #2 –
It still amazes me, how people keep posting about "how to save space on hdd" (given these times of ultra-cheap storage)
Can't wait for a HD-AAC encoder :P

how much does stereo "cost" ?

Reply #3 –
I've found that when I have questions like yours the only satisfactory answer is to create test files (e.g. stereo vs mono) of the audio I typically listen to and then look at the file sizes.

30 minutes of my own experiments usually gives me more peace of mind than reading online discussions.  I recommend you start creating different test files and seeing for yourself what the cost is.

how much does stereo "cost" ?

Reply #4 –
Quote from: odyssey on
It still amazes me, how people keep posting about "how to save space on hdd" (given these times of ultra-cheap storage)


Well, portable players still have pretty low capacities as manufacturers have essentially stopped using hard drive and are now relying on solid state.  I don't really think that 8GB, 16GB, 32GB, or even 64GB of storage is a whole lot especially if someone has a 13,000 song library.  There is also an overwhelming large amount of people who continue to store media solely on their computer's internal primary hard drive.

Not everyone realizes that they can spend $120 and get a 2TB external drive for backing their content up.

how much does stereo "cost" ?

Reply #5 –
Quote from: chrizoo on
As a side-question: if the talkradio broadcasts I mentioned basically consist of one guy speaking and sitting in front of a stereo mic, what difference is there actually between a mono and a stereo recording? My guess is that if the person moves his head from left of the mic to the right, you would notice this movement reflected in the levels of the L and R speakers. But if we assume the person doesn't change his position at all and is centered in front of the mic, there should not be any noticeable difference between a mono and stereo recording, right? And if the speaker sits somehow on the left of the mic, both speakers would basically emit the same signal, but at a lower level on the R speaker and that's all to the story ....

.... OR is there supposed to be a spatial effect by which I mean (for a lack of a better description) you can somehow hear/feel like you were in the radio studio or hear/feel like the radio host were in your room while losing this effect completely in mono ?


Most likely he would have recorded with a single mic, and thus there would be no L-R spatial information.  A single mic can only pick up one channel.  A reason you may want to keep it in stereo is for any music they used.

BTW, another factor with what Kornchild said is that nowadays, many people aren't just using their players for music.  Video and even games are using up space, so I can still see a desire for compressed music (at least on the go).

how much does stereo "cost" ?

Reply #6 –
Quote from: odyssey on
It still amazes me, how people keep posting about "how to save space on hdd" (given these times of ultra-cheap storage)

well, I'd be thankful if you could accept the premiss. People may have their reasons. And different priorities than you.
Kornchild mentioned solid state HDDs. That's one. Also consider the additional time needed for daily backups plus twice the space for backup drives.
And just because you consider storage cheap doesn't mean it makes much sense to waste space if it is not necessary.

Quote from: Drexl on
A reason you may want to keep it in stereo is for any music they used.

true, but in my case there is no music

Quote from: Drexl on
Most likely he would have recorded with a single mic, and thus there would be no L-R spatial information.  A single mic can only pick up one channel.

http://www.google.com/search?&q=%22stereo+microphone%22
I use them myself.

Quote from: odigg on
I recommend you start creating different test files and seeing for yourself what the cost is.
If you read my posting, you'd know that I did that.
The problem is how to measure. See A) and B)

how much does stereo "cost" ?

Reply #7 –
One other thing you might have to deal with is the issue of compatibility.

I know of a few devices and apps that play MP3 which have issues with mono files.  They expect all files to be stereo, and any mono-encoded file will be played back improperly.

I do a lot of MP3 encodings of talk radio, especially come American football season. I record all the radio pre-game and post-game broadcasts directly off-air.  I pipe the radio's line-out to my soundcard's line-in and set the radio output volume and line-in volume such that replaygain settings are around +1.0 dB and peak of around 0.5

I record them to standard CD format 44 kHz/16-bit stereo WAVs, one hour at a time, one file per hour.

I then encode to LAME MP3 V5 VBR with a lowpass of 10000 Hz.  It usually produces a bitrate of ~90 kbps or about 40 MB per hour and the full 3 hours often fits under 128 MB of space.

That lowpass eliminates nearly all the background static, since it's only people speaking 90% of the time and normal human speech doesn't reach 10 kHz.  Any music is usually from commercials and isn't important.

how much does stereo "cost" ?

Reply #8 –
Quote from: chrizoo on
Quote from: odyssey on
It still amazes me, how people keep posting about "how to save space on hdd" (given these times of ultra-cheap storage)

well, I'd be thankful if you could accept the premiss. People may have their reasons. And different priorities than you.
Kornchild mentioned solid state HDDs. That's one. Also consider the additional time needed for daily backups plus twice the space for backup drives.
And just because you consider storage cheap doesn't mean it makes much sense to waste space if it is not necessary.

You ask about differences between mono and stereo lossy-encoded - Even with an extremely large library, the difference will most likely turn out to be smaller than you think given that most efficiently encoded mp3s uses joint stereo. You'll save a few pennies and thats it. I'm just trying to save you from wasting a LOT of your time with something that doesn't give you very much in the end.
Can't wait for a HD-AAC encoder :P

how much does stereo "cost" ?

Reply #9 –
Quote from: chrizoo on
As a side-question: if the talkradio broadcasts I mentioned basically consist of one guy speaking and sitting in front of a stereo mic, what difference is there actually between a mono and a stereo recording? My guess is that if the person moves his head from left of the mic to the right, you would notice this movement reflected in the levels of the L and R speakers. But if we assume the person doesn't change his position at all and is centered in front of the mic, there should not be any noticeable difference between a mono and stereo recording, right? And if the speaker sits somehow on the left of the mic, both speakers would basically emit the same signal, but at a lower level on the R speaker and that's all to the story ....

.... OR is there supposed to be a spatial effect by which I mean (for a lack of a better description) you can somehow hear/feel like you were in the radio studio or hear/feel like the radio host were in your room while losing this effect completely in mono ?


When someone talks into a microphone, the microphone also picks up reflections from the room and furniture around the mic. To make mono = stereo there woulld have to be at the very least perfect symmetry of the entire room around the center line of the microphones.

how much does stereo "cost" ?

Reply #10 –
Quote from: odyssey on
I'm just trying to save you from wasting a LOT of your time with something that doesn't give you very much in the end.

That's kind of you.
well the idea is to find out how to measure how much it is. If it turns out to be pennies, I can decide accordingly. But if you look at the example in my OP, it doesn't quite look like pennies. The mono file needs 60% of the stereo file's size ... that's only half the size almost!

My OP poses a methodology question, but answers were on a different level so far.

how much does stereo "cost" ?

Reply #11 –
No doubt there are compatibility issues somewhere but I’ve made a great many mono mp3s of speech without running into any problems so far. Joint stereo, in my tests, results in a larger file, even if for no reason other than dither. I can’t foresee equipment or program circumstances getting worse to make me regret my approach.

I make mono, mostly by selecting one of the two channels that exist in almost all my sources. Anything important not the same in both channels is rare to non-existent. I resample to 22.05kHz. I encode with LAME V8, preventing it from resampling to 16kHz, or whatever it’s default, and from cutting off below 11kHz.

The result is easy to understand and provides comfortable listening. Any differences from the source are difficult to detect, although some people probably hear differently. VBR size is unpredictable, but I’ve put almost 50 hours on one CD-R.

how much does stereo "cost" ?

Reply #12 –
Thank you, AndyH-ha. What's the rationale for not averaging the two channels (but selecting one instead) ?
50 hours? That's only about 14 MB per hour. Or ~30 kbps and still "any differences from the source are difficult to detect". Not bad.
Can you post your command line params?

how much does stereo "cost" ?

Reply #13 –
Quote from: Arnold B. Krueger on
When someone talks into a microphone, the microphone also picks up reflections from the room and furniture around the mic. To make mono = stereo there woulld have to be at the very least perfect symmetry of the entire room around the center line of the microphones.
Thank you Arnold. Funny, I had the same thought.
But I errouneously thought that in practice, if the radio anchor speaks directly into his microphone, the reflections you mentioned affect the overall accoustics, but don't have a big difference on L/R differences. Wrong. The reason for my error is that I mistakenly picked up a file which I thought to be stereo but turned out to be a dual-track mono file (identical L/R tracks). And since I - obviously - couldn't hear any differences between this file and its mono downmix I erroneously concluded that assuming the person always keeps the exact position in front of the mic, mono would sound just like stereo.

So please ignore my posting #2.
It's useless, my mistake.

(OP still stands though).

how much does stereo "cost" ?

Reply #14 –
My sources are mostly either CDs or cassettes. My daughter buys expensive book tapes and I make back-ups. Mixing tape channels works fine generally, but sometimes the duplicators are not well balanced and the “identical” channels are not very. CDs don’t have this problem, but both sources are mono in every one of probably a hundred books (some multi-book series) I’ve done, except for a bit of music at the beginning or end on many books. This being stereo, or mixed to mono from stereo, doesn’t interest me. I’d just as soon it disappeared).

Given this, mixing to mono, rather than selecting one channel, is just an extra, useless, step to make the job take longer. I do all processing, excepting the mp3 encoding, in an audio editor. LAME resampling isn’t very clean and I don’t choose to use it.

The encoding line is -V 8 --vbr-new --resample 22 --lowpass 11 --noreplaygain. There may be some changes for later versions of LAME, but the 3.97 I’m using isn’t broken, so I don’t fix it.

how much does stereo "cost" ?

Reply #15 –
Quote from: odyssey on
It still amazes me, how people keep posting about "how to save space on hdd" (given these times of ultra-cheap storage)


Quote from: kornchild2002 on
Not everyone realizes that they can spend $120 and get a 2TB external drive for backing their content up.


@both, ultra-cheap is a very subjective term ... not only 120 dollars has not the same value for all the world, but on a lot of counties, HDDs are much more expensive, in dollars, and in local currency, than in USA/Europe... besides, if hdd is so cheap, why people care about the 1% increase in compression that ALAC gives and such?

how much does stereo "cost" ?

Reply #16 –
Quote from: odyssey on
It still amazes me, how people keep posting about "how to save space on hdd" (given these times of ultra-cheap storage)


While local storage is cheap, the online world is full of amazingly slow links and servers. For example, every week I upload for public consumption a ca. 30 minute MP3 file of a lecture. The server I upload to restricts file transfers to such a slow rate that it simply makes my life easier and to code this file at the lowest rate possible consistent with acceptable intelligibility.  Using .wav or flac files aren't even an option.

The source material is well-represented in mono, and the instant reduction in data rate that comes with coding it in mono really works. Furthermore, I  know that managing the bandwidth requirements for effective listening to this file makes it more acessible to the people who want to listen to it in real time.

I've managed to filll several TB with my audio and video production work in the past few years. When you're out of space hwenyou're doing a project, its a real PITA right now!. Even with 3 TB of new drives in the next room, it can still take a lot of work and time to perform an effective upgrade.

how much does stereo "cost" ?

Reply #17 –
Quote from: AndyH-ha on
cassettes


I'm sorry, cassettes?

Which source in which industrialized country offers cassettes in 2010? I haven't seen a common cassette tape since 1995.

how much does stereo "cost" ?

Reply #18 –
Quote from: AndyH-ha on
Given this, mixing to mono, rather than selecting one channel, is just an extra, useless, step to make the job take longer. I do all processing, excepting the mp3 encoding, in an audio editor. LAME resampling isn’t very clean and I don’t choose to use it.

The encoding line is -V 8 --vbr-new --resample 22 --lowpass 11 --noreplaygain. There may be some changes for later versions of LAME, but the 3.97 I’m using isn’t broken, so I don’t fix it.

I am confused.

how much does stereo "cost" ?

Reply #19 –
One thing you should consider, a woman's voice while speaking is only 2 kHz and a soprano  might hit a 4 kHz note.  This means talk radio broadcasts can withstand massive compression.  Check the table posted in this site's lame wiki that lists the pitch cutoffs at different V settings.  I would then listen to the resulting audio.  I found lame produces too many artifacts below 40 BR for my liking.  If size is really an issue you might explore other encoders.  Audio books I have downloaded from the internet library have been as low as 16 BR and sound just fine.

I would think with all that compression, you might not worry about the stereo/mono option. 

I do not consider myself to be an authority, but I think AM radio is only 50 BR, FM 70 BR and HD 90 BR.  If I am correct, you might not want to capture radio as lossless.

how much does stereo "cost" ?

Reply #20 –
@doccolini: the --resampler parameter of LAME can do two things: resample, and .... not resample.

LAME is tuned in such a way that depending on the quality that you choose, it resamples to a lower samplerate than the original file. Using the --resample parameter in such cases allows to tell the encoder to keep the original samplerate.

So, in that case, Andy is using files sampled at 22Khz (as he has explained above) and forcing LAME not to resample.

how much does stereo "cost" ?

Reply #21 –
Some people may not be aware that some bread is made from whole wheat rather than refined flour -- if they never pay attention.

how much does stereo "cost" ?

Reply #22 –
Since I am particular about how I process any audio that I want to keep, regardless of what other people may think would be better, I would probably never "capture" at other than loseless, regardless of the probably quality of the source. One can always start over and do things a bit differently if they retain the unprocessed source until convinced that the final product is exactly as desired. I might, however, record at a lower sample rate, I might record as mono if the source is mono (although it is generally safer to capture two channels when they exist, just in case one has a problem).

One of my soundcards, which does a very good job at 44.1kHz and higher sample rates, produces a rather strange result at lower sample rates: it is best to test whatever you decide to do before you start on the real data.

how much does stereo "cost" ?

Reply #23 –
Quote from: [JAZ] on

So, in that case, Andy is using files sampled at 22Khz (as he has explained above) and forcing LAME not to resample.

Ah... Ok then, I didn't read carefully.

how much does stereo "cost" ?

Reply #24 –
Quote from: dhromed on
Quote from: AndyH-ha on
cassettes

I'm sorry, cassettes?

Which source in which industrialized country offers cassettes in 2010? I haven't seen a common cassette tape since 1995.

I recently found a very old radio drama (used, on Amazon) that I listened to as a child and has been out of print for nearly a decade. It was never released on CD. I'm still trying to find a way to digitize this so I can add it to my archive. But this is off-topic.