Ogg Vorbis problems

Please can Vorbis users do the quick test at the end and report results in this thread. I don’t care about MP3, we know it works, just vorbis, flac and others that aren’t the standard MP3.

Ogg Vorbis playback has screwed the pooch. This is also a problem on the Nokia N900 super phone (oops, smart phone / *nix pocket PC)

provided the post isn’t sanitized the following is a bit of research done by someone else:

tuomas.kulve.fi/blog/2009/11/07/n900-battery-duration-ogg-vs-mp3/

Search ‘n900 nokia libvorbis ffmpeg’ under your favourite brand search engine and it should be the first hit.

This is pretty promising as it states that using the libvorbis decoder you lose about 40% battery life, using ffmpegs vorbis decoder you gain about 2% (lets call it equal with mp3 for the sake of argument).

It being promising is based on Sandisk being hoopy froods who really know where their towels are at and get their fingers out, suck some nourishment, and fix the decoder.  And, if that doesn’t make sense, well never mind.

I’ve just done a test against Meat Loaf Bat 1, Bat 2, and the mild insult to Jim Steinman - Bat 3.

Vorbis encoded to 192 VBR (q6)

MP3 encoded to 128 VBR (v5)

MP3 encoded to 192 VBR (v2)

Playing the MP3 128 kb/s tracks (admittedly just the tracks in alphabet order as the playlists & tags were AWOL) I used a whole 2% battery after 9 tracks (~40 mins playback). Taking the units is dead at 50% into account this does actually equate to ~15hr playback (adjust 2% to 4 - 5 % battery use).

Playing the Vorbis back I lose about 2% per track (~6 mins) which comes to around 3 hours (I’m getting about 4.5hrs total playback time in previous straight runs). So this actually tallies. Yep, my whole collection is in vorbis, and I ain’t switching to MP3. Na, ah. They support vorbis on the tin, so they can properly support it in the device.

I already have a complaint / whining about product session open with Sandisk so this will be passed on to them. The tech details of this are also on a battery issue thread.

I’ve had a quick look, but shall provide a proper playtime report on the weekend - right now it’s silly o’clock and I want sleep. Ain’t going to find it, but nothing to stop me wanting.

Please encode a CD or two in ogg and mp3 - preferable both 128 and 192 kb/s for each, however, I’m not exactly in a position to make demands. Completely different CDs between the two formats, although not scientific, will provide some info if that’s what you have. 

Play back each format in turn and provide the battery use after ~1hr / 1 CD. Sticking to the following format should make comparison simple.

Format: MP3

DRM: Yes / No (preferable not, but could provide interesting info for drm’d users)

Bit rate:

VBR/CBR:

Time played:

Battery % Before:

Battery % After:

(optionally) album name:

Format: Ogg

Bit rate:

VBR/CBR:

Time played:

Battery % Before:

Battery % After:

(optionally) album name:

Thanks to those who help out here, good and bad results wanted.

All tracks are non-DRM (evil stuff that DRM), driving into a load of 4 ohm speakers. Test will be repeated with supplied earphones and Sennheiser 64 ohm earphones. Album is Bat Out of Hell with a few bonus tracks.

As can be seen the low impedance speakers cause heavy battery drain (why not a higher impedance buffer amp at the front? - strange Logitech hey?).
Increasing bitrate makes no noticeable difference to battery life. However, the drain of the speakers will mask small changes hard to find. (9% battery use for each MP3 set, 16% for Vorbis)

Format: MP3
Bit rate: 128
VBR/CBR: VBR
Time played: 3976 (66:16)
Battery % Before: 100%
Battery % After: 91%

Format: MP3
Bit rate: 192
VBR/CBR: VBR
Time played: 3976 (66:16)
Battery % Before: 91%
Battery % After: 82%

Format: Ogg Vorbis
Bit rate: 192
VBR/CBR: VBR
Time played: 3976 (66:16)
Battery % Before: 82%
Battery % After: 66%

You can’t really test by looking at the before and after battery percentages, since they extremely inaccurate.  You need to run the battery down for each format and time how long it takes (recording the output on a PC is an easy way to do that), or else use a DMM to measure power consumption directly.

However, people have done this test before, and the Sandisk version of Tremor is quite slow.  This is partially Sandisk’s fault for not optimizing the codec much, and partially Tremor’s fault for not incorporating more optimizations by other groups.  Eventually I want to put the Rockbox Tremor optimizations back into main Tremor so commercial devices decode Vorbis efficiently, but its a low priority for me since its a lot of work and i tend not to use them much. 


You can’t really test by looking at the before and after battery percentages, since they extremely inaccurate.  You need to run the battery down for each format and time how long it takes (recording the output on a PC is an easy way to do that), or else use a DMM to measure power consumption directly.


Quite true, it’s meant nothing more than a rough guide. But, when you get near double the drain showing, the results are pretty conclusive.

I looked for previous info, didn’t find anything - I wouldn’t have bothered with this otherwise. It was just a warning (and hopefully a confirmation from others) that the advertised support for Vorbis is complete rubbish. If the products battery life is around a third for one of the formats due to poor implementation of the decoder, you can’t advertise ‘support’ for the format (well, fair enough if they had made the problem clear on the box).

 “partially Sandisk’s fault” ? No, it’s entirely Sandisk’s fault (from a customer’s point of view). If they want to buy (hell, they ain’t even paying for it here) a piece of software and proceed to sell on to the end user it is entirely their responsibility to make sure it’s up to scratch. I can’t go to the Tremor developers and request a refund for a faulty product, I can go to Sandisk under warranty or the retailer.

When designing electronics I don’t throw in any old component that fits the rough description, I look for something that does the job properly - if it doesn’t exist on the market, I design in the discrete equivalent or get a contractor to do it. Either way, the end product is wholly my responsibility and I make damned sure it works.

Further to this, if this is a common problem, why is Sandisk Customer ‘Support’ taking two days to reply to every e-mail, and feeding me crap about it ‘just being a one off problem and I should get a request unit’? Frankly, they are an offensive bunch of thieves who want to use us as nothing but cash cows to be milked dry. Another supplier I won’t be touching from now on. It sickens me that this list is now so long.

Thanks for the info you have provided, please don’t think the above is aimed at you. It’s a rant at the sick state most modern companies are in.

“Eventually I want to put the Rockbox Tremor optimizations back into main Tremor so commercial devices decode Vorbis efficiently”

Sounds like you are an open source developer? For this I salute you and thank you. One of the last small pockets of people who are still out there trying to better things and doing it for no better reason than that ‘they can’.

dihenydd wrote:

 

 

“Eventually I want to put the Rockbox Tremor optimizations back into main Tremor so commercial devices decode Vorbis efficiently”

Sounds like you are an open source developer? For this I salute you and thank you. One of the last small pockets of people who are still out there trying to better things and doing it for no better reason than that ‘they can’.

Yes, saratoga is one of the Rockbox developers.

So, unless it is supported by Rockbox I won’t find a player that is capable of supporting Ogg Vorbis files then?

And by support, I’m emphasing the word support, not ‘play for a couple of hours’.

Or am I mistaken? Hell, I’d even stoop to Sony or Apple if they were to support Vorbis correctly.

I’m looking for:

MSC (I’m a linux user, and no - I’m not installing anything to manage the device. I’ve written my own scripts for managing my library the way I want it).

Playlist

Tags

Music no video (I’d obviously settle for video on the player if all else is good, I will just never use it)

Happiness

Vorbis

Battery life >12 hrs (50 hour life, down to 20 for Vorbis playback is fine).

Am I asking for too much in life?

I can’t believe Vorbis gets half the battery life of  similar bitrate mp3s. Perhaps it might be 25% less battery life, although others claim only a 10-15% decrease in battery life. I’ll stick with 256 kbps mp3 CBR for this reason, as even variable bitrate mp3 probably gets shorter battery life than CBR. While I could store 33% more music on the player using mp3 VBR averaging 192 kbps rather than 256 CBR, and I might get almost the same sound quality, I probably won’t get the same battery life though. The scarcest thing on the Clip+ is battery life.

JK98 wrote:
  I’ll stick with 256 kbps mp3 CBR for this reason, as even variable bitrate mp3 probably gets shorter battery life than CBR. While I could store 33% more music on the player using mp3 VBR averaging 192 kbps rather than 256 CBR, and I might get almost the same sound quality, I probably won’t get the same battery life though.

I’m not sure I agree with your theory. The higher the bitrate, the more battery used…why else would the ratings always be based on 128kbps? I’ve ripped a few albums at V2 LAME, which is the 192 average, and only an extremely complex passage in a song would go over 220kbps, much less 256. I think you should test it out…take one of your players, put a V2 album on repeat, and see how long it plays, then repeat the test with a 256CBR. CBR is a waste of space, this test could determine how it affects battery life.

http://forums.sandisk.com/sansa/board/message?board.id=clip&thread.id=14326

(Remembering that the Clipv2 and Clip+ have nearly identical hardware) 

“CBR is a waste of space, this test could determine how it affects battery life.”

A shortage of space is not the problem, the problem is a shortage of battery life. My old one GB player got 18 hours on one AAA battery. With that player, battery life was not the problem, storage space was. So I went from a player with a battery slot and unlimited battery life with spare batteries to a player with a card slot and unlimited storage with spare cards, but a built in battery! UGH! Why can’t I have both swappable memory and swappable battery in the same player, and get good versatility?