Help the clueless - mp3 quality.

[commando]

Quote:

Originally Posted by breez Are you serious? 600kbps to 1100kbps is about 75-140 kilobytes per second. 10 year old hardware can manage that easily off the HD

I said "especially if your PC's already busy or old". My work PC is brand new, but I work in software development and I need all the power I can get. Playing back a FLAC encoded song can increase my build time from 20 minutes to 22 minutes.

 

[Mr.Radar]

If you have DMA enabled for your hard drive and no DSPs or dithering in Foobar then FLAC should actually use less CPU time than most lossy formats because of it's integer decoder, especially on older computers.

 

[commando]

For my situation, hard drive access is more important than CPU. I use a resampler and the EQ in foobar too, but it only ever uses about 5% of the CPU.

 

[hackeron]

Quote:

Originally Posted by commando I said "especially if your PC's already busy or old". My work PC is brand new, but I work in software development and I need all the power I can get. Playing back a FLAC encoded song can increase my build time from 20 minutes to 22 minutes.

Now that more and more hardware players support FLAC, its sure to decrease the battery life as well.

But FLAC eats quite a bit of CPU along side keeps hard drive busy. I dont know how it is on portable players, but on a laptop it wouldnt take any more power due to increased hard drive use, but due to increased decoding cpu time. Is this issue applicable to higher bitrate mp3s? -- I'll check, it definitely applies to whatever format the ipod uses.

EDIT: flac is quicker to decode than lossy? -- really? --- I heard its quite a bit slower to decode. I'll look at this further.

Oh, and on a 400mhz busy laptop, ps aux doesnt show mp3 decoding to take more than about 2% cpu (granted I use a highly efficient decoder).

 

[commando]

Quote:

Originally Posted by hackeron Now that more and more hardware players support FLAC, its sure to decrease the battery life as well. But FLAC eats quite a bit of CPU along side keeps hard drive busy. I dont know how it is on portable players, but on a laptop it wouldnt take any more power due to increased hard drive use, but due to increased decoding time. Is this issue applicable to higher bitrate mp3s? -- I'll check, it definitely applies to whatever format the ipod uses. EDIT: flac is quicker to decode than lossy? -- really? --- I heard its quite a bit slower to decode. I'll look at this further.

Foobar2000 is playing back flac on my PC now, a 2.8GHz hyperthreaded P4, and it's only taking about 2% of the CPU. That's nothing to me. The disk access is more important for what I do with my PC.

I'd never use FLAC on a portable device, even one with a hard drive. What happens is the device spins up the hard drive, copies the music to RAM, then plays it. If you're using mp3s it might do this even 30 minutes. If you use flac it'd be more like every 10 minutes. That's going to kill the battery life.

I have no idea about the decoding time, myself.

 

[commando]

hackeron, you seem to edit your posts quite a lot right after you first make them, so replying to them gets a bit difficult. Have a look at the post I quoted compared with the one that's there now. Can I suggest using the preview button? I don't mean to be rude, it's just a suggestion, you might get better answers to your questions/discussion that way.

 

[hackeron]

Quote:

Originally Posted by commando hackeron, you seem to edit your posts quite a lot right after you first make them, so replying to them gets a bit difficult. Have a look at the post I quoted compared with the one that's there now. Can I suggest using the preview button? I don't mean to be rude, it's just a suggestion, you might get better answers to your questions/discussion that way.

Its just you reply so damn quick, lol.

 

[hackeron]

Quote:

Originally Posted by commando Foobar2000 is playing back flac on my PC now, a 2.8GHz hyperthreaded P4, and it's only taking about 2% of the CPU. That's nothing to me. The disk access is more important for what I do with my PC. I'd never use FLAC on a portable device, even one with a hard drive. What happens is the device spins up the hard drive, copies the music to RAM, then plays it. If you're using mp3s it might do this even 30 minutes. If you use flac it'd be more like every 10 minutes. That's going to kill the battery life. I have no idea about the decoding time, myself.

Are you sure?, From my understanding, how the players are made is to preload part of a playing MP3 into ram, some even preload the entire mp3 (while some dont preload at all).

If they indeed spin down the hard drive, then it would have to spin back up every 3-4 minutes and spinning up the hard drive causes more tare and takes about 5-10 times more power than to have the hard drive just spin.

New portable hard drives take a flat 2-3W power when they spin, but 20W for up to 30 seconds when you spin them up.

Maybe some players have some clever preloading of multiple MP3s by predicting shuffle and what not, but think about it, if you manually select track, it would have to remove all the preloaded ones from memory and spin up the hard drive. Hard drive takes up to 10 seconds to respond which is a major usability factor.

Maybe I'm wrong in all of this, but do these portable players actually come with that much ram?

The flash memory ones sure dont, and flash memory takes a lot more power to access than ram which at the battery size differences, would explain the battery lifes.

If portable hard drive players indeed preload up to 30 minutes of music, the player should last atleast, what 80 hours with the included battery? -- Certainly atleast as much as flash memory players concidering how much more it takes to access flash memory compared to ram.

And portable players come with a 5 times bigger battery and half the battery life, something is fishy.

 

[Mr.Radar]

Quote:

Originally Posted by hackeron EDIT: flac is quicker to decode than lossy? -- really? --- I heard its quite a bit slower to decode. I'll look at this further.

I just ran a test (all the lossy files are sourced from the FLAC file) and here are the results:

FLAC (-6 I think, or maybe -7, 749kbps): 2172 ms to decode
Ogg Vorbis (Q5, 149kbps): 2896 ms to decode
LAME (Alt Preset Insane, 320 kbps): 2495 ms to decode
LAME (Alt Preset Standard, 181 kbps): 1995 ms to decode

It seems that lower bitrate MP3's win out, however FLAC is no slouch either. Ogg Vorbis took the most time to decode, despite having the lowest average bitrate of the bunch.

Also, I must comment that LAME takes FOREVER to encode files.

EDIT: BTW, the file used in the test was Telephone Line by ELO (track 2 off their "An New World Record" CD, mine says mastered in 1992 on it).

 

[hackeron]

Quote:

Originally Posted by Mr.Radar I just ran a test (all the lossy files are sourced from the FLAC file) and here are the results: FLAC (-6 I think, or maybe -7, 749kbps): 2172 ms to decode Ogg Vorbis (Q5, 149kbps): 2896 ms to decode LAME (Alt Preset Insane, 320 kbps): 2495 ms to decode LAME (Alt Preset Standard, 181 kbps): 1995 ms to decode It seems that lower bitrate MP3's win out, however FLAC is no slouch either. Ogg Vorbis took the most time to decode, despite having the lowest average bitrate of the bunch. Also, I must comment that LAME takes FOREVER to encode files.

Thats hardly a fair test.

If anything you should have all of them use the same bitrate (except for flac obviously).. Also, what software?

I'm going to run a test myself now, and lets see if I just happen to find a completely opposite conclusion

Now, where was that Chopin cd....

 

[Mr.Radar]

How is that test unfair? I used Foobar 2000 with the Speed Meter (for Diskwriter) plugin. My system is an AMD AthlonXP 2500+ with 512 megs of RAM and a Seagate 7200.7 (w/ 8 megs of cache) UltraDMA 100 HDD.

EDIT: I should also mention that the times I list were the average of three runs for each file.

 

[hackeron]

Quote:

Originally Posted by Mr.Radar How is that test unfair? I used Foobar 2000 with the Speed Meter (for Diskwriter) plugin. My system is an AMD AthlonXP 2500+ with 512 megs of RAM and a Seagate 7200.7 (w/ 8 megs of cache) UltraDMA 100 HDD.

You used different bitrates and you didnt use a native decoder.

 

[Mr.Radar]

Quote:

Originally Posted by hackeron You used different bitrates

Most people who use lossy encoding (and know what they're doing) try to get the best sound quality for the bitrate so it isn't unfair to use different bitrates. BTW, you yourself wondered how bitrate affected the decoding speed and I provided results (low-bitrate MP3 did win, which goes against what I said earlier in this thread). My HDD is fast enough that the bitrate shouldn't have any impact.

Quote:

and you didnt use a native decoder.

Umm... what? I can see what you mean, but I a) don't have any tools for measuring the speed on Windows and b) Foobar2000 is an "equalizer" for the results; all the audio is processed exactly the same, decoded at the same priority, etc. With the individual decoders they can change settings like priority that can affect the outcomes in favor of one codec or another.

I have to go eat dinner. I'll be back in 10-15 mins.

 

[hackeron]

Quote:

Originally Posted by Mr.Radar Most people who use lossy encoding (and know what they're doing) try to get the best sound quality for the bitrate so it isn't unfair to use different bitrates. BTW, you yourself wondered how bitrate affected the decoding speed and I provided results (low-bitrate MP3 did win, which goes against what I said earlier in this thread). My HDD is fast enough that the bitrate shouldn't have any impact. Umm... what? I can see what you mean, but I a) don't have any tools for measuring the speed on Windows and b) Foobar2000 is an "equalizer" for the results; all the audio is processed exactly the same, decoded at the same priority, etc. With the individual decoders they can change settings like priority that can affect the outcomes in favor of one codec or another. I have to go eat dinner. I'll be back in 10-15 mins.

FIrst of all, increased bitrate with the same format, ofcourse. -- Testing with different formats is not a fair test.

Also, thats where I come in

-- I dont use that poor excuse for a GUI, so results are comming.

 

[commando]

Quote:

Originally Posted by hackeron Are you sure?, From my understanding, how the players are made is to preload part of a playing MP3 into ram, some even preload the entire mp3 (while some dont preload at all). If they indeed spin down the hard drive, then it would have to spin back up every 3-4 minutes and spinning up the hard drive causes more tare and takes about 5-10 times more power than to have the hard drive just spin. New portable hard drives take a flat 2-3W power when they spin, but 20W for up to 30 seconds when you spin them up. Maybe some players have some clever preloading of multiple MP3s by predicting shuffle and what not, but think about it, if you manually select track, it would have to remove all the preloaded ones from memory and spin up the hard drive. Hard drive takes up to 10 seconds to respond which is a major usability factor. Maybe I'm wrong in all of this, but do these portable players actually come with that much ram? The flash memory ones sure dont, and flash memory takes a lot more power to access than ram which at the battery size differences, would explain the battery lifes. If portable hard drive players indeed preload up to 30 minutes of music, the player should last atleast, what 80 hours with the included battery? -- Certainly atleast as much as flash memory players concidering how much more it takes to access flash memory compared to ram. And portable players come with a 5 times bigger battery and half the battery life, something is fishy.

If you read my post a little more carefully you'll see that i've described preloading of songs on a hard disk into memory on the player. I think this is a common problem with you posts - lack of reading/understanding what people have said, and the constant editing of your posts shows that you post first and think second. I don't mean that to be offensive, just a suggestion about how to get along here better.

Yes, hdd players know the order of the playlist, even on random, so it can preload the next few songs into ram. They do spin the drive down while it's not needed. They can have 64MB of ram or so, enough for a lot of music at 128kbps. I'd be very suprised if any portable player currently on the market doesn't do this caching. If the user manually skips songs the cache will probably be emptied and whatever the user has selected is loaded. That's why you get better battery life if you use playlists rather than jumping randomly around your music.

Flash is a type of RAM, you don't need to cache it.