Need help for choosing a (probably Rockbox-supported) player

[sargeborsch]

Hello. I'm considering buying a new player, which should either be supported by Rockbox, or have all the features of Rockbox "out of the box", and also good audio quality and power. (not worse/less than my current player — Cowon V5, which has 30mW per channel, if I understand it right)
It must support real gapless playback and 24-bit WavPack hybrid playback (without crude conversion to 16-bit or something like this.)

And good things to have (but I don't believe it's possible) are ReplayGain (non-destructively make all tracks/albums sound with the same perceived loudness, without clipping, adjustable target gain) and Crossfeed DSP (compensate for excess channel separation that headphones have, something like http://wiki.hydrogenaud.io/index.php?title=Foobar2000:Components/Crossfeed_%28foo_dsp_crossfeed%29) — they will probably make me free from having to pre-process all music before putting it into player.

Please, give me some hints about what to look at.

Thank you for your attention. And I'm sorry for my English, it's not my native language yet…

[gomezz]

I have just replaced my aging Creative Zen V Plus but decided to go for the equally venerable Sansa Clip+ (*) with which I am very happy apart from two aspects:  a) It is all too easy to accidentally turn it on by touching the ON/OFF button so I put it in a padded pouch when not in use in my bag;  b) I need to be able to find and use the Play/Pause button by touch while wearing rubberised cotton work gloves so I glued a bit of plastic to the button to create a noticable bump (using the tip of the clip off an old biro pen).  Otherwise it is just about perfect for my needs when Rockboxed and with the addition of a 32GB SD card to add to the 8GB internal memory is more than enough to hold my whole CD collection in mp3 format twice over.

(*) I did look at the newer Sansa Clip Zip model but that appears to be not as well sorted a device.

[saratoga]

Cowon V5, which has 30mW per channel, if I understand it right)

FWIW that really doesn't mean anything.

It must support real gapless playback and 24-bit WavPack hybrid playback (without crude conversion to 16-bit or something like this.)

Probably no such device exists.  But it doesn't matter.  There is no real advantage to 16 vs. 24 bit for portable devices.

And good things to have (but I don't believe it's possible) are ReplayGain (non-destructively make all tracks/albums sound with the same perceived loudness, without clipping, adjustable target gain) and Crossfeed DSP (compensate for excess channel separation that headphones have, something like http://wiki.hydrogenaud.io/index.php?title=Foobar2000:Components/Crossfeed_%28foo_dsp_crossfeed%29) — they will probably make me free from having to pre-process all music before putting it into player.

Clip with rockbox?  Cheap and easy to try out. 

[sargeborsch]

>There is no real advantage to 16 vs. 24 bit for portable devices

When in noisy environment — it's definitely true. But when it's quiet "outside", that's a nice thing to have… and I tend to use player in both very noisy conditions (while riding bike in the city) and in quiet conditions.

>FWIW that really doesn't mean anything.

Then how do I measure the maximum loudness? I am pretty sure it's different for different devices, as previously I had one or two problematic players which were unable to give enough loudness, and became totally useless in noisy conditions.

>Clip with rockbox?  Cheap and easy to try out. 

You mean "Sansa Clip+"?

[saratoga]

You'll basically never be able to get the noise floor low enough that quantization error matters, and if you did, we have noise shipping anyway.

The power figure is basically meaningless for a lot of reasons. First because you need to know an impedance to interpret those numbers. Second because the actual max power is determined by the supply voltages programmed into the firmware making the nominal values irrelevant. They just put those numbers out because they expect people to not know what they mean.

Are you actually running your player at max volume?  If so getting louder headphones might be a better idea that looking for single db differences between players.

[sargeborsch]

Yes, sometimes I run at max volume (sometimes when riding bicycle), but I have some headroom in most of the recordings (unless they have "unusually" high dynamic range) — I use ReplayGain normalization when encode music for portable device, and also reduce higher frequencies (EQ). My headphones are already pretty loud (Audio-technica ATH-M50), and I doubt there are any other headphones that are louder and at the same time have not worse sound quality (FR, distortion). And if the "quiet" tracks would require more dynamic compression to be heard in noisy conditions, this wouldn't be cool.

OK, if the numbers are meaningless, then how do I get an idea of how much "power" do they produce (relatively)? that is, which player can be louder when given the same recording.

>You'll basically never be able to get the noise floor low enough that quantization error matters, and if you did, we have noise shipping anyway

maybe you are right here.
if doing postprocessing and then encoding as WavPack hybrid, choosing 16 bit dithered vs 24 bit not dithered hurts compression efficiency a bit, but this is very tiny, maybe impossible to notice any difference by ears…

[saratoga]

Yes, sometimes I run at max volume (sometimes when riding bicycle), but I have some headroom in most of the recordings (unless they have "unusually" high dynamic range) — I use ReplayGain normalization when encode music for portable device, and also reduce higher frequencies (EQ). My headphones are already pretty loud (Audio-technica ATH-M50), and I doubt there are any other headphones that are louder and at the same time have not worse sound quality (FR, distortion). And if the "quiet" tracks would require more dynamic compression to be heard in noisy conditions, this wouldn't be cool.

You can actually go quite a lot louder than those headphones, or alternatively, use an external amp.  They're of moderate to low sensitivity, so the output voltage of a portable device may not drive them super loud. 

OK, if the numbers are meaningless, then how do I get an idea of how much "power" do they produce (relatively)? that is, which player can be louder when given the same recording.

You'd compare the output impedance and full scale voltage and use that with the published sensitivity of your headphones to compute the SPL.  However, manufacturers rarely give this information out, so often it must be measured. 

Fortunately there is little difference in volume between most portable devices, with almost everything putting out about 0.4-0.7 Vrms. 

if doing postprocessing and then encoding as WavPack hybrid, choosing 16 bit dithered vs 24 bit not dithered hurts compression efficiency a bit, but this is very tiny, maybe impossible to notice any difference by ears…

16 bit dithered compresses worse than 24 bit?   What are these files you're encoding?  Are they actually 24 bit?

[sargeborsch]

>16 bit dithered compresses worse than 24 bit?   What are these files you're encoding?  Are they actually 24 bit?

Yes, they are (technically) even true 32 bit floating point, because they are output from several non-trivial DSPs (crossfeed, EQ, ReplayGain compensation), so even if the source is 16 bit, after processing it's not and cannot be losslessly stored as 16 bit.
If I wouldn't need to do this, then most of my music can be encoded as 16 bit without any drawbacks, because it's already 16 bit and thus no additional quantization/dithering would be needed.

About headphones. Maybe they are not the most sensitive, but they already can reveal various noises/hum in unchancy devices, such as laptops' on-board sound cards. (Or this is a different characteristic?) And, anyway, I don't want to change them. Carrying an external amp is too inconvenient, too. I hoped that it's not that hard to find a player which can deliver at least the same loudness as Cowon V5. But if it is hard, I'd better keep all components as they are… they do the job, just not very conveniently.

[saratoga]

>16 bit dithered compresses worse than 24 bit?   What are these files you're encoding?  Are they actually 24 bit?

Yes, they are (technically) even true 32 bit floating point, because they are output from several non-trivial DSPs (crossfeed, EQ, ReplayGain compensation), so even if the source is 16 bit, after processing it's not and cannot be losslessly stored as 16 bit.

I suspected you were doing something like that if the 24 bit files compressed better.  If you're just taking 16 bit audio and upconverting it, I wouldn't do that.  For things like EQ/Replaygain/crossfeed you can just use rockbox anyway.

[sargeborsch]

Well, my current player is not supported. That's why I'm asking for recommendations.

[404_user_not_found]

I like sansa clip+ with rockbox support, but I can't use the sansa clip+ forever and I have a trouble for choosing a player. It's not limited to the lossy formats that player can support. Most new players uses a database and doesn't have an option to use a simple folder playing style and don't have a hotkey to quickly skip a folder. Some player significantly slow downs when you have over 9000 tracks and more. I need a player, that can be used blindly, folder playing style, SD card support and ability to quickly skip a folder. Any recommendations?

[gomezz]

The Sansa Clip+ plays lossless FLAC files so not sure why you are saying this is a limitation of this device?