Previous day | Jump to hour: 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 | Next day

Seconds: Show Hide | Joins: Show Hide | View raw
Font: Serif Sans-Serif Monospace | Size: Small Medium Large

Click in the nick column to highlight everything a person has said.
The Logo icon identifies that the person is a core developer (has commit access).

#rockbox log for 2023-11-29

00:10:51 Join jacobk [0] (
00:25:40 Quit othello7 (Ping timeout: 276 seconds)
00:32:26 Quit jacobk (Ping timeout: 245 seconds)
00:44:35 Join jacobk [0] (~quassel@
00:56:07 Quit LjL (Read error: Connection reset by peer)
00:56:39 Join LjL [0] (~ljl@user/ljl)
01:18:11***Saving seen data "./dancer.seen"
01:18:17 Quit pixelma (Quit: .)
01:18:17 Quit amiconn (Quit: - Chat comfortably. Anywhere.)
01:19:15 Join pixelma [0] (
01:19:15 Join amiconn [0] (
01:21:15 Quit davisr (Remote host closed the connection)
01:25:47 Quit ddevault (Changing host)
01:25:47 Join ddevault [0] (e7b4bb6755@sourcehut/staff/ddevault)
01:36:30 Join IPG [0] (~InvoxiPla@
01:59:36 Join aaabbb [0] (
02:39:24 Join termac [0] (~termac@user/termac)
02:39:42 Quit jacobk (Ping timeout: 268 seconds)
02:51:21aaabbbhow active is rockbox development?
03:18:15***Saving seen data "./dancer.seen"
03:24:38termacthere are people working on it
03:38:53termaci got a question regarding the hifi walker h2, not related to rockbox itself: when i connect my headphones i have a static background noise and i need to set the volume very low like 10/100 or lower
03:40:37termacis there something on the player i can adjust to make it work better with these headphones or do i need to add extra resistance between the player and the headphones
03:42:07termacearphones are these:
03:54:26 Join tm512 [0] (~tm512@user/tm512)
03:57:41 Join ac_laptop [0] (~ac_laptop@2001:910:107e:1:e29d:31ff:fe2d:a258)
03:58:33 Join jacobk [0] (~quassel@
04:16:27tm512kinda seems like the Sansa devices are some of the best rockbox-capable devices on a budget? been thinking of picking up an MP3 player to play around with, just don't want to spend that much on something that does only a subset of what my phone already does
04:17:24jssfrI'm pretty happy with the sansa family, yes
04:17:45jssfraudio quality is decent, they have physical buttons, a microsd slot, and that's good in my book.
04:17:49tm512I have quite a bit of nostalgia for MP3 players though, was in high school when they were still massively popular, probably more common than smartphones
04:19:56 Quit jacobk (Ping timeout: 245 seconds)
04:20:26tm512what's the battery situation like on these older Sansa devices nowadays?
04:20:57jssfrholds up for several days of occasional use, certainly for 10+ hours
04:21:10jssfralso, I have a USB-C->A adapter for my phone, which allows me to charge the sansa from the phone :-)
04:23:03tm512a quick google search brought up this and the use of papyrus as a font on the packaging certainly inspires confidence /s
04:23:50jssfrthe batteries are also soldered, which is ... not fun
04:23:56jssfreasy to short while unsoldering the wires
04:23:58jssfrat least in some models
04:26:11*tm512 does not even have a soldering setup
04:26:38tm512maybe that's something I should change
04:31:27tm512looks like the NOMAD Jukebox ZEN has replacable batteries and if I'm not mistaken, 2.5" HDDs are basically non-flash CompactFlash cards. seems like the perfect device, though it doesn't look like rockbox supports those?
04:32:59tm512or wait, not 2.5", the 1.3" ones or whatever
04:34:44tm512wikipedia says those use 2.5" drives but those devices don't look *that* large. but I dunno
04:34:48jssfr1.3" HDDs are hard to come by *and* you cannot just swap an SSD in
04:34:58 Quit CH23_M (Ping timeout: 256 seconds)
04:35:02jssfrI did that with my iriver and it fried the power regulation system because the SSD needed more amps
04:35:10jssfrmy iriver also didn't like a CF adapter unfortunately
04:35:32 Join CH23_M [0] (~CH23@revspace/participant/ch23)
04:37:58tm512does seem like the 1.3" HDDs are the exact same form factor as a CF card, both using an IDE interface as well. dunno about the power consumption. the thought of flash memory drawing more power than an HDD seems odd to me
04:38:09jssfroh it's not odd
04:38:14jssfrI've got several instances of that, in fact.
04:38:45jssfrI have three 2.5" disks in my home lab: a HDD (500mA peak), and two SSDs (1A and 1.5A peak)
04:38:58jssfrflash needs high currents for write operations IIRC
04:39:21jssfrwhile HDDs only need high currents for spinup, and that can be brought down by spinning up at a lower rate, which low-RPM drives (as commonly found in small form factors) generally do.
04:42:07tm512are you including CF cards as SSDs? I know they're all flash memory, and "solid state", but I thought the type of flash could differ drastically, with memory cards typically being a sort that is way less durable
04:42:34 Join jacobk [0] (~quassel@
04:43:15jssfrCF cards didn't work at all in my device so I haven't tried it.
04:43:36jssfrI guess for CF you can find the maximum current numbers in some specification and compare that to the numbers on the HDD you aim to replace.
04:47:48tm512not sure which devices I'd even consider that use a microdrive and support rockbox, the iPod Classics seem on the pricier end
04:48:01tm512at least the newer ones, like 5G and up were what I looked at
04:48:41tm512like easily >$100 whereas I could get a Sansa for under $50 and get a microSD slot with the latter
04:50:29tm512I guess I'd just worry about the battery when/if it becomes an issue, maybe send it and a new battery to someone with actual soldering skills to perform the replacement
04:50:50jssfrsounds reasonable
05:10:01tm512another thing I was curious about. do codecs make a significant difference on battery life on most players? like the impression I've gotten is that musepack has about the best compromise between compression efficiency and computational complexity, though if it doesn't matter much I'd probably just use opus in cases where I'm making my own encodes from lossless
05:11:50jssfrI do know that opus (used to be) more complex than e.g. vorbis; while my iriver was able to decode vorbis in realtime without clock frequency boosts, opus required those.
05:12:15jssfrI'm not sure what the current situation is, I assume that there has been some optimization.
05:12:52aaabbbthe libopus decoder has significantly improved
05:12:59aaabbbbut it's still heavier than mp3
05:17:07tm512assuming an SoC that can handle it, would it make much of a difference on battery life over a cheaper codec?
05:18:16***Saving seen data "./dancer.seen"
05:22:51aaabbbis this solid state or spinning?
05:24:32tm512the Sansa e200 series (which I'm looking at) is all flash
05:25:32aaabbbso mp3 decoding is much more efficient on that than musepack
05:27:34aaabbbat least on your hardware
05:28:54 Quit CH23_M (Ping timeout: 260 seconds)
05:29:24tm512aaabbb: at 96kbit it seems MP3 has a big advantage but it looks like they get closer at higher bitrates
05:30:46tm512I'm not sure exactly what the MHz values on this table is actually representing
05:31:23aaabbbthat's clock frequency needed for realtime play
05:31:35jssfr(theoretically, I guess)
05:31:43aaabbbyeah, roughly
05:31:48jssfrbecause I doubt that the clip can go to ~1 GHz :D
05:32:19aaabbbwell it means that's what it would have to go to in order to achieve realtime
05:32:38aaabbbwhether the hardware can hit the required frequency is another question... haha
05:32:57jssfrI mean, briefly it probably can.
05:32:58tm512the v1 e200 apparently has two 80MHz ARM7TDMI cores
05:33:13tm512and the v2 has a 250MHz ARM9TDMI
05:33:51aaabbbjssfr: i doubt it would even recognize a 1ghz signal as a clock signal
05:34:14aaabbbtm512: which version do you have?
05:34:23aaabbband does rockbox have any way to utilize the two core anyway?
05:35:18tm512aaabbb: I don't have one (yet)
05:37:39tm512I was thinking of picking one up. I dunno if I care much whether it's the older ARM7 one or the ARM9. seems anything other than HE-AAC would be handled just fine on the ARM7
05:38:46aaabbbarm9 has tcm. idk if rockbox makes use of the tcm, but if it does, it should allow significant performance improvements for codecs that don't require much memory
05:38:51aaabbbIF it use the tcm
05:38:55tm512though these results are quite old, 2010, dunno if the situation has changed, but I don't think I have any HE-AAC stuff laying around, nor do I have much of a reason to make copies with that codec
05:39:53aaabbbxHE-AAC isi a wonderful codec, slightly more efficient than opus at low bitrates
05:40:02aaabbbnot regular HE-AAC tho :)
05:48:21tm512looking at these results though it seems like musepack would be a reasonable choice for cases where I'm not starting with a mid-low bitrate source
05:49:17aaabbbsince they're so close at mid-high bitrates, i'd care more about the audio quality of those codecs
05:52:52 Quit termac (Ping timeout: 255 seconds)
05:54:43 Join termac [0] (~termac@user/termac)
05:55:52tm512mpc is supposed to be transparent around 160-192kbit but I guess even at 128kbit, it gives vorbis a run for its money. opus would still be better, if it doesn't noticeably affect battery life
05:56:28aaabbbopus is something like 30% heavier than vorbis if i remember
05:56:31aaabbbit used to be over double
06:04:12tm512I'll have to do some ABX tests on myself, because if I end up being unable to reliably tell the difference between opus and mpc at the bitrates I plan on using, even with my headphone setup, then I'm better off just using the cheaper codec
06:04:46tm512certainly wouldn't be able to tell the difference with the Sansa's DAC and headphones that I actually feel comfortable taking out of the house
06:05:32aaabbbjust because you can't ABX one sample doesn't mean you can ABX another. idk how mpc works but with opus, it may be nearly transparent, but then suddenly go into intensity stereo which you may n otice
06:06:33tm512also it's a bummer these devices don't have bluetooth support, because nowadays using a bluetooth speaker is perhaps the most common use I have for mobile audio playback
06:06:54 Quit sch (Ping timeout: 268 seconds)
06:07:11aaabbbbt reduces audio quality because it reencodes to a different lossy format, so there's some generation loss
06:11:58tm512I don't think it really matters considering the quality of the bluetooth speaker I have, which if I had to be kind to it all I could say is "it's better than not having music"
06:19:55 Quit termac (Ping timeout: 276 seconds)
06:19:58tm512I figure if that use case really matters with this device I'd just pick up a bluetooth transmitter. though passing the analog output from the MP3 player into a cheap ADC for further lossy encoding is not ideal especially if I got a better speaker or decent wireless headphones
06:23:32tm512I could, perhaps, just look for a portable speaker with an analog input
06:36:47 Join tricky [0] (~tricky@2600:6c48:503f:4550::9019)
07:18:17***Saving seen data "./dancer.seen"
07:25:08 Quit tricky (Quit: Leaving)
09:18:18***Saving seen data "./dancer.seen"
10:02:34 Join dconrad_web [0] (~dconrad_w@
10:06:15dconrad_webtermac: the h2 definitely has a relatively hot output, and those se215s look to be pretty sensitive. I have noticed a small amount pf background noise on my player when using sensitive IEMs like that in a very quiet room, but nothing that you would notice with even quiet audio playing
10:08:39dconrad_webif I recall correctly, the noise doesn't scale with volume though - you may want to try with some different headphones and see if those are better?
10:10:05dconrad_webcould also try an attenuator too, there used to be one specifically intended for iems, but I'm not sure if its still for sale. I seem to recall it was kind of expensive for what it was too
10:26:07 Quit dconrad_web (Quit: Connection closed)
10:32:51 Join othello7 [0] (
10:48:58 Quit Maxdamantus (Ping timeout: 255 seconds)
10:49:38 Join Maxdamantus [0] (~Maxdamant@user/maxdamantus)
11:18:19***Saving seen data "./dancer.seen"
11:18:29speachytm512, jssfr: FWIW, the iriver devices work with CF just fine, but you need the latest bootloader and a nightly build newer than the 3.15 release.
11:18:44speachythe OF doesn't work at all with CF cards though.
11:20:14speachythe high write currents of "flash" devices are heavily skewed by mSATA and nvme stuff that has high write speeds. CF cards are much lower power, and SD cards are lower yet
11:21:09speachy(And consider that the spin-up current is awful, and applies wthether it's a read or write operation)
11:22:35 Join sch [0] (
11:24:46speachyaaabbb: on a typical port, TCM/IRAM is used for the most performance critical code, and yes, that includes parts of some of the codecs. At least the ones that we've tuned, anyway.
11:27:27speachyalso, wrt codecs and their theoretical quality, consider that the analog stage of most of these DAPs isn't all that great, and unless you have golden headphones (and ears) you'd be hard pressed to tell the difference in the real world. Also, IME the actual CPU core is only a minority of the overall powre consumption. The analog path (poer supplies, headphone amp, clocks/PLLs that are always
11:27:29speachyrunning whenever something is playing back.. that's the main power draw.
11:27:56speachy(storage and the screen also tend to completely dwarf the actual CPU)
11:28:42speachya while back I implemetned dynamic reclocking for the jz47xx targets, from ~540MHz down to under 200. The difference wasn't measurable in the battery benchmark.
11:39:53speachy(ie on the order of a minute or two across a >12 hour test...)
12:09:51 Join davisr [0] (~davisr@fsf/emeritus/davisr)
13:18:20***No seen item changed, no save performed.
13:41:46 Quit jacobk (Ping timeout: 255 seconds)
13:48:05 Join lebellium [0] (
13:51:43 Join termac [0] (~termac@user/termac)
13:53:41 Join jacobk [0] (~quassel@
13:59:27 Quit davisr (Remote host closed the connection)
14:00:04 Join davisr [0] (~davisr@fsf/emeritus/davisr)
15:18:24***Saving seen data "./dancer.seen"
16:01:52 Quit jacobk (Quit: No Ping reply in 180 seconds.)
16:03:30 Join jacobk [0] (~quassel@
16:47:18tm512speachy: wow I would have expected reclocking to make some kind of difference at least. I guess I'll probably end up opting for opus
16:54:26speachyolder stuff is more power hungry at the CPU core level, but yeah, that surprised me too.
16:54:44tm512quality of a codec is only a minor consideration. I'd take a "worse" codec for improved battery life, since regardless of what I pick I'll probably be using similar bitrates, though opus will probably provide more headroom for maintaining transparency even with difficult sections of music
16:55:50tm512like I'd probably use opus at 160kbit or 192kbit, unless my listening tests show some inadequacy there, but I kinda doubt it
16:56:16speachyopus is pretty hard to beat honestly.
16:56:53 Quit Malinux (Ping timeout: 240 seconds)
16:57:10tm512I spent almost $500 on a headphone setup (DAC/amp + headphones) years ago and to my ears it was only a moderate improvement over my previous headphones going through onboard DACs/amps
16:57:16speachymp3 is still quite good at ~128k, and low CPU load. most of the argument for higher compresson codecs is to use less disk space/bandwidth but that's honestly not much of an issue
16:57:17tm512I definitely don't have golden ears
16:57:46speachyhah, yeah, and even if you had golden years once, age is relentless. :D
16:59:04tm512I was honestly surprised by LAME @ 128kbit from a lossless source. totally inconsistent with my memories of 128kbit MP3s from the 00s
17:00:31tm512I wasn't using my headphones for that quick listening test, but the only codecs that performed terribly at 128kbit were MP2 and WMA (through ffmpeg's encoders)
17:00:32speachyearly mp3 encoders definitely lacked versus more modern ones. IIRC LAME started as a series of patches against the reference encoder.
17:01:45tm512even MP2 wasn't that offensive, but both versions of WMA that ffmpeg can encode had those characteristic artifacts that I always thought sounded like the music was playing underwater
17:01:58speachyhaving more CPU cycles to throw at the compression helps a lot. keep in mind that all of these are *decoder* specifications; different encoders can take different approaches to achieve the same output waveform
17:04:09tm512my experience with 128kbit music in the past probably involved quite a bit of generation loss, like ripping a CD that a friend burned that was already sourced from MP3s just downloaded online, which might've already been reencoded from a lossy source
17:04:31speachyyeah that's gonna be suboptimal no matter what
17:18:27***Saving seen data "./dancer.seen"
17:23:33 Join Moriar [0] (
17:40:28 Quit ac_laptop (Ping timeout: 276 seconds)
17:52:22 Quit jacobk (Ping timeout: 268 seconds)
17:54:06 Quit lebellium (Quit: Leaving)
17:56:14tm512though picking up a Sansa is probably my best option it is tempting to get one of these iRiver devices to have more 68k stuff in my collection
17:56:36tm512seems all of the supported iRiver players use a coldfire SoC?
17:56:44speachythe m68k stuff is probably the most underpowered of any current device.
17:57:04speachyI think the H10 family uses the same SoC as the older ipods.
17:57:43speachyI do miss my old H120 though. A very solid unit.
17:58:53tm512going for 68k wouldn't really be a practical consideration, I'm just partial to the architecture, lol
18:00:50tm512I've got three 68000 machines right here along with a 68040 that basically self-destructed (capacitors, of course)
18:09:35 Quit termac (Ping timeout: 264 seconds)
18:20:13tm512can the Sansa e200R be either v1 or v2, or is it only ever one of those? not sure if it's still the case but the wiki suggests that the v2 has worse battery life than the v1
18:20:49tm512well specifically following the link to
18:49:24 Join jacobk [0] (~quassel@
18:52:14tm512so it seems like the Rhapsody models have the v1 (ARM7TDMI) SoC?
19:18:31***Saving seen data "./dancer.seen"
19:38:23 Join ac_laptop [0] (~ac_laptop@2001:910:107e:1:e29d:31ff:fe2d:a258)
19:51:40 Quit ac_laptop (Quit: WeeChat 3.8)
21:02:51 Quit davisr (Remote host closed the connection)
21:18:32***Saving seen data "./dancer.seen"
21:22:31aaabbbspeachy: i think the quality of the codec matters when you're trying to get the lowest bitrate that is transparent with your equipment and ears. the dac might have a crappy noise floor but if the highs from 128kbps mp3 are cut off, but 64kbps opus sounds transparent, then that's a clear win for opus
21:27:18aaabbbtm512: opus 160 or 192 is way overkill
21:29:16aaabbb96kbps is almost always transparent for most people, except classical music where sometimes you need 128kbps
21:29:52aaabbband unless you have very good equipment or get bothered by occasional dips into intensity stereo, 64kbps is also fine
21:35:19tm512aaabbb: well, I'll do my own listening tests. I'd want the minimum that's reliably transparent to my ears, plus maybe 32kbps as headroom
21:36:02aaabbbi think you'll be very impressed with it :)
21:36:25aaabbbhigher bitrate will reduce battery life slightly tho
21:37:22tm512with a Sansa player, having the option to use a microSDHC card, storage isn't a big issue. a 32GB card is like $5 and is big enough to not necessitate going down to 96kbit or lower
21:39:24tm512at least with my desktop speakers, 64kbit opus does perform admirably. haven't checked with my headphone setup though
21:39:45tm512I feel like I can tell there's a difference but this isn't a blind test, so *shrug*
21:40:08tm512at worst, it's pretty slight
21:40:10aaabbbyou'll probably notice 64k not being transparent with good headphones
21:40:29aaabbbbut 96k will certainly be transparent for most samples
21:41:28tm512track I just did this non-blind test with is "Crown Shy" by Plaid
21:41:46tm512gonna have to set up this:
21:42:16tm512unless there's better FOSS ABX software, this is just what I found with a quick search
21:42:31aaabbbhydrogenaudio lists some foss abx software
21:42:36aaabbbalso you can make your own with a bash script
22:29:07tm512still not a blind test, but yeah honestly 96kbit sounds great to me. friend was able to distinguish it from lossless though, apparently
22:29:39aaabbba bash script could be a blind test if it chooses the samples at random
22:30:12tm512ended up linking her to a set of 3 flac files, with the original, 64kbit opus, and 96kbit opus. apparently I forgot to strip the album art metadata from the original though
22:30:41tm512she said she thought it was a red herring so I dunno if it actually spoiled the results
22:31:00aaabbbit's useless if it's not blind. you can distinguish 96kbit from lossless in a few rare situations like with symbols, certain kinds of classical music or something with a lot of detail in the side channel
22:36:19 Quit massiveH (Quit: Leaving)
22:36:25tm512I'd probably end up targeting 128kbit, then, assuming a proper ABX test proves I can't tell the difference on multiple different tracks from different genres
22:37:04tm512seems promising though, really good codec
22:47:59 Quit othello7 (Ping timeout: 264 seconds)
23:01:35 Quit jacobk (Ping timeout: 260 seconds)
23:12:26 Join jacobk [0] (~quassel@
23:18:36***Saving seen data "./dancer.seen"
23:37:17 Quit m01 (Quit: Konversation terminated.)
23:38:52 Quit sch (Ping timeout: 246 seconds)
23:40:08 Join m01 [0] (

Previous day | Next day