phofman said:
Hi phofman.
I guess you misunderstood, what I actually meant.
I am saying that many ( the majority?)DACs (not the DAC - chips itself!) do have an ASRC in front of the DAC chip, which upsamples any incoming samplerate stream usually to 96 resp. 192khz.
What I am saying is that In this case it's worth to try upsampling on the PC.
Or even better - playing 96/192 upsampled material.
Cheers
soundcheck said:
I am not sure majority of external DACs feature ASRC, I would dare saying this feature is still rather rare. I did not get the feeling LaKrissen is using one of these. On the other hand, most DAC chips used in external DACs as well as sound cards feature the controlled upsampling I talked about, except for the old NOS DACs used in DIY.
In any case, upsampling to the final conversion frequency using quality algorithms in the PC makes sense, be it integer multiple or straight 192kHz, depending on the actual DAC.
phofman:
Highest quality DACs such as
Benchmark DAC1, ESS Sabre, EMU 0404 USB ( to name some if the best out there)
using ASRC.
Highest quality DACs such as
Benchmark DAC1, ESS Sabre, EMU 0404 USB ( to name some if the best out there)
using ASRC.
You are right, the highest quality external DACs are using ASRC. Probably not a majority of DACs though.
Are you sure EMU 0404 USB is using ASRC? I have not seen a sound card with ASRC yet. What advantage would ASRC bring if the card can control pace of its incoming data directly, such as the asynchronous USB mode in EMU 0404 USB?
Are you sure EMU 0404 USB is using ASRC? I have not seen a sound card with ASRC yet. What advantage would ASRC bring if the card can control pace of its incoming data directly, such as the asynchronous USB mode in EMU 0404 USB?
Re: 24bit/192khz dac output form laptop
Well you are absolutely right .... no new information is invented
however the information there is, gets delivered much clearer into my ear through the DAC in my laptop with the higher resolution, 16/44.1 sounds harsh and bloated when compared with 24/192 which sounds so much clearer .... and firmer
i guess the DAC is just crappy at 16/44.1
at least the difference is HUGE in 24/192
try comparing 128 mp3 to a 320 mp3 of the the same piece of music... that's about the difference i get
Best regards LaKrissen
gainphile said:
Well you are absolutely right .... no new information is invented
however the information there is, gets delivered much clearer into my ear through the DAC in my laptop with the higher resolution, 16/44.1 sounds harsh and bloated when compared with 24/192 which sounds so much clearer .... and firmer
i guess the DAC is just crappy at 16/44.1
at least the difference is HUGE in 24/192
try comparing 128 mp3 to a 320 mp3 of the the same piece of music... that's about the difference i get
Best regards LaKrissen
Which sound card is best for Linux audio?
Hi everybody. Sorry for the (partial) off-topic. This is a rather collateral (yet fundamental!
) question.
Eventually I'm going to build a new PC. I plan to use it as a digital source (transport) for my Hi-End system as well as for HT (5+1, Full HD video, etc). I will only use digital I/O for HiFi stereo, but maybe I'll use the card analog outputs for multichannel (HT). Needless to say, it will run Linux.
Now the question is: which audio card should I buy for it?
Of course, it must be able to synchronize its output stream to an incoming one and/or to an external clock (i.e. the one coming from the DAC... 😉 ).
And of course it must be perfectly and fully supported by ALSA.
That is, the question is: which audio card (possibly not costing a fortune! ) is currently the best one for Hi-End audio use with Linux?
I was thinking about an Emu 1212m, but from what I have seen on the ALSA WiKi it is still only partly supported (and possibly buggy, too).
A next guess could be an ESI Juli@, but I've read discordant opinions about it... 😕 would it be good enough? are there better options?
Maybe some M-Audio?
What else?
P.S.: PCI support in MB is fading out... are there any suitable PCI-express based card?
Please give your advice!
Hi everybody. Sorry for the (partial) off-topic. This is a rather collateral (yet fundamental!
) question.Eventually I'm going to build a new PC. I plan to use it as a digital source (transport) for my Hi-End system as well as for HT (5+1, Full HD video, etc). I will only use digital I/O for HiFi stereo, but maybe I'll use the card analog outputs for multichannel (HT). Needless to say, it will run Linux.
Now the question is: which audio card should I buy for it?Of course, it must be able to synchronize its output stream to an incoming one and/or to an external clock (i.e. the one coming from the DAC... 😉 ).
And of course it must be perfectly and fully supported by ALSA.
That is, the question is: which audio card (possibly not costing a fortune!
) is currently the best one for Hi-End audio use with Linux?I was thinking about an Emu 1212m, but from what I have seen on the ALSA WiKi it is still only partly supported (and possibly buggy, too).
A next guess could be an ESI Juli@, but I've read discordant opinions about it... 😕 would it be good enough? are there better options?
Maybe some M-Audio?
What else?
P.S.: PCI support in MB is fading out... are there any suitable PCI-express based card?
Please give your advice!
motherboard and USB soundcard comparison
Iv'e been comparing 2 different soundcards on my PC. One is on the motherboard and the other is USB.
Intel ICH5 - IEC958: which is on the motherboard and has an optical SPDIF out. This connects into a Behringer DEQ2496. I haven't done anything to the sample rate and the DEQ2496 lists the source as being 48kHz. There is no noise carried over from the PC and it sounds great. I use this with XMMS.
The other sound card is a Poppulse USB which uses the C-Media. I connect it to my DEQ2496 via an XLR AES3. The DEQ2496, using this with Amarok and plughw:1,0 lists the source as being 44.1kHZ. It is also noiseless and sounds great.
If I switch between the 2 inputs using the DEQ2496 input selector and play the same source from XMMS and Amarok at the sametime there really isn't much difference between the two. I'm pretty certain that a listener doing blind tests would have a hard time discerning between the two.
would anyone like to suggest any possible improvements?
The DEQ2496 output connects via an optical SPDIF to a Citypulse DA7.2x DAC
col.
Iv'e been comparing 2 different soundcards on my PC. One is on the motherboard and the other is USB.
Intel ICH5 - IEC958: which is on the motherboard and has an optical SPDIF out. This connects into a Behringer DEQ2496. I haven't done anything to the sample rate and the DEQ2496 lists the source as being 48kHz. There is no noise carried over from the PC and it sounds great. I use this with XMMS.
The other sound card is a Poppulse USB which uses the C-Media. I connect it to my DEQ2496 via an XLR AES3. The DEQ2496, using this with Amarok and plughw:1,0 lists the source as being 44.1kHZ. It is also noiseless and sounds great.
If I switch between the 2 inputs using the DEQ2496 input selector and play the same source from XMMS and Amarok at the sametime there really isn't much difference between the two. I'm pretty certain that a listener doing blind tests would have a hard time discerning between the two.
would anyone like to suggest any possible improvements?
The DEQ2496 output connects via an optical SPDIF to a Citypulse DA7.2x DAC
col.
RME HDSP series is one of the best supported professional PCI cards under Linux.
With the extension board it can be extended to 6*analog out.
Cheers
With the extension board it can be extended to 6*analog out.
Cheers
I had a RME Digi96 (predecessor of Hammerfall), perfect linux support, synchronizes to an external clock... if your gear takes ADAT, it's an excellent bet.
It's called oversampling, not upsampling. Upsampling is a marketing buzzword.
Anyway, yes, the oversampling done with enough bits of precision (like, 64 bits floating point) WILL sound better than the oversampling done in a cheap DAC IC where they tried to trim the number of gates to the absolute minimum, losing bits in the process, and adding roundoff errors... no surprise here. It's possible to do it right in an IC... it just costs a few extra cents...
Well, there are two ways of doing ASRC :
- as in the ESS Sabre (the simple and correct way) : oversample to a few megahertz and then use linear interpolation
- as in all other chips : having a low output sample rate (like, 192k, not 10 MHz) makes it a lot more complex, and the errors introduced are harder to eliminate...
It's called oversampling, not upsampling. Upsampling is a marketing buzzword.
Anyway, yes, the oversampling done with enough bits of precision (like, 64 bits floating point) WILL sound better than the oversampling done in a cheap DAC IC where they tried to trim the number of gates to the absolute minimum, losing bits in the process, and adding roundoff errors... no surprise here. It's possible to do it right in an IC... it just costs a few extra cents...
phofman said:
Well, there are two ways of doing ASRC :
- as in the ESS Sabre (the simple and correct way) : oversample to a few megahertz and then use linear interpolation
- as in all other chips : having a low output sample rate (like, 192k, not 10 MHz) makes it a lot more complex, and the errors introduced are harder to eliminate...
soundcheck said:
yeah, that would be nice... unfortunately they are far too expensive for my current budget. 🙁
(though maybe not so important, AFAIK they are also limited to 96K... while it would be nice to be able to go all the way up to 192K).
peufeu said:
no, my current DAC (North Star model 192) does not take ADAT. Only S/PDIF (either optical or coax), AES/EBU and a custom I2S input.
For the future I'm also considering the Sabre (I have to find someone that have built it in my area to do some comparison with my current DAC...).
Though AFAIK the Sabre (at least in its TP incarnation, but I'm not willing to design my own PCBs...) doesn't take ADAT either.
yup... though (in marketing) the word "upsamplig" is usually used to indicate some "smarter" oversampling techniques than just adding 0s between samples before applying digital filtering. Quite often it also implies some form of requantization (e.g. 16 to 24 bit conversion) and/or ASRC.
To my knowledge all the audio cards with envy24* are well supported in linux - alsa.
I have a TerraTec Aureon 7.1 Space with spdif (toslink) in and out and i am able to sync this card to a blank spdif signal i send to the spdif in.
With alsamixer i can choose various internal clock or external from spdif in.
ciao
andrea
I have a TerraTec Aureon 7.1 Space with spdif (toslink) in and out and i am able to sync this card to a blank spdif signal i send to the spdif in.
With alsamixer i can choose various internal clock or external from spdif in.
ciao
andrea
anbello said:
mmmh... so we have:
AudioTrak
Prodigy 7.1 HiFi (?)
Prodigy 192
Prodigy HD2 (?)
ESI
Juli@
M-Audio
Audiophile 192,
Revolution 7.1 (?)
Onkyo
SE-200PCI (hard to find, expensive...)
TerraTec
Aureon 5.1 (max 48KHz/16 bit ?!)
Aureon 7.1 (is this still available?)
Anything else?
What about an Asus Xonar D2X?
Surely an RME HDSPe AIO looks so much better than anything else... if it wasn't so expensive! 🙁
(BTW: is it supported? ALSA Matrix does not list it explicitly 😕 ).
soundcheck said:
what would you choose between an AudioTrak Prodigy 7.1 HiFi and ASUS XONAR D2X?
...or should I try to find some 2nd hand RME?
Re: Which sound card is best for Linux audio?
You are right, support for EMU is far from complete.
It has only stereo outputs, not enough for the intended multichannel output. On the other hand, it can output true 176.4kHz SPDIF, unlike most other ICE1724-based cards which output 88.2kHz SPDIF in 176.4kHz internal clock mode (seems like a bug in the chip to me).
Several ICE1724-based cards offer SPDIF input via a common MI/ODI/O add-on card. Note that it is fully supported/tested only for Prodigy192 (since I did not have another combination available 🙂 )
I have not seen a PCI-e card with separate crystal clock for 44.1kHz and 48kHz fs families. The better ones have 48kHz straight, 44.1kHz using PLL, the worse ones use
PLL for both cases. I would expect the PLL clock has more jitter, but I have not done any measurements or comparison listening tests.
I think ASUS XONAR D2X has a single crystal too, see http://www.ixbt.com/multimedia/asus/d2/chips2b.jpg .
I would go for cards with SPDIF output transformer (usually a black block close to the SPDIF-out cinch, easily discernible on googled pictures). IIRC, regular TOSLINK can go only to 96kHz.
RME are splendid cards, the question is whether you can actually utilize their potential aimed at multitrack music creation.
UnixMan said:
You are right, support for EMU is far from complete.
It has only stereo outputs, not enough for the intended multichannel output. On the other hand, it can output true 176.4kHz SPDIF, unlike most other ICE1724-based cards which output 88.2kHz SPDIF in 176.4kHz internal clock mode (seems like a bug in the chip to me).
Several ICE1724-based cards offer SPDIF input via a common MI/ODI/O add-on card. Note that it is fully supported/tested only for Prodigy192 (since I did not have another combination available 🙂 )
I have not seen a PCI-e card with separate crystal clock for 44.1kHz and 48kHz fs families. The better ones have 48kHz straight, 44.1kHz using PLL, the worse ones use
PLL for both cases. I would expect the PLL clock has more jitter, but I have not done any measurements or comparison listening tests.
I think ASUS XONAR D2X has a single crystal too, see http://www.ixbt.com/multimedia/asus/d2/chips2b.jpg .
I would go for cards with SPDIF output transformer (usually a black block close to the SPDIF-out cinch, easily discernible on googled pictures). IIRC, regular TOSLINK can go only to 96kHz.
RME are splendid cards, the question is whether you can actually utilize their potential aimed at multitrack music creation.
Re: Re: Which sound card is best for Linux audio?
any idea about the Audiotrak Prodigy 7.1 HiFi ?
Can it output true 176.4kHz SPDIF?
Does it have the separate crystal clock for 44.1kHz and 48kHz fs families?
Does it have SPDIF output transformer(s)? (from the only small picture I have seen maybe... but I'm not sure).
How 'bout its clock jitter?
unfortunately that link doesn't work (anymore?). 🙁 But I trust you. 😀
AFAIK standard S/PDIF coax too, for that matter. It was S/PDIF in itself to be limited to 24/96 (I guess no one was dreaming of anymore back then...). Though it have been somehow "extended" (in a more or less de-facto standard way) to work up to 24/192 too.
of course not... most likely I will not use it for recording at all.
...but on the other end I'm quite intrigued by their ultra-low jitter clocks! (they claim jitter < 1ps !)
any idea about the Audiotrak Prodigy 7.1 HiFi ?
Can it output true 176.4kHz SPDIF?
Does it have the separate crystal clock for 44.1kHz and 48kHz fs families?
Does it have SPDIF output transformer(s)? (from the only small picture I have seen maybe... but I'm not sure).
How 'bout its clock jitter?
unfortunately that link doesn't work (anymore?). 🙁 But I trust you. 😀
AFAIK standard S/PDIF coax too, for that matter. It was S/PDIF in itself to be limited to 24/96 (I guess no one was dreaming of anymore back then...). Though it have been somehow "extended" (in a more or less de-facto standard way) to work up to 24/192 too.
of course not... most likely I will not use it for recording at all.
...but on the other end I'm quite intrigued by their ultra-low jitter clocks! (they claim jitter < 1ps !)
Re: Re: Re: Which sound card is best for Linux audio?
Take a look at google images. It has two crystals around the PCI controller chip (as all ice1724-based cards do).
Transformer - take a look at this image http://www.mp3car.com/vbulletin/att...-review-audiotrak-prodigy-7-1-hifi-opamps.jpg . Do you see the silk-screen diagram of the transformer, without the actual component installed? I do not think this particular version has it.
I have not tested it, but unless it is using some upgraded version of Envy24HT, I doubt it outputs 176.4kHz SPDIF when clocked from on-board crystals.
Clock jitter - crystals, supply regulators, PCB layout, ... - I cannot tell without any measurements. I would guess better than PLL-clock cards.
Google is your friend 🙂 http://forum.purepc.pl/Audio-f63/Asus-Xonar-D2-t241101.html
I vaguely remember I could not convince TOSLINK connection to HK AVR to transfer more than 96kHz. But my memory leaks a lot 🙂
I very doubt that figure in the super-noisy PC environment. But no hard data to support my doubts.
UnixMan said:
Take a look at google images. It has two crystals around the PCI controller chip (as all ice1724-based cards do).
Transformer - take a look at this image http://www.mp3car.com/vbulletin/att...-review-audiotrak-prodigy-7-1-hifi-opamps.jpg . Do you see the silk-screen diagram of the transformer, without the actual component installed? I do not think this particular version has it.
I have not tested it, but unless it is using some upgraded version of Envy24HT, I doubt it outputs 176.4kHz SPDIF when clocked from on-board crystals.
Clock jitter - crystals, supply regulators, PCB layout, ... - I cannot tell without any measurements. I would guess better than PLL-clock cards.
Google is your friend 🙂 http://forum.purepc.pl/Audio-f63/Asus-Xonar-D2-t241101.html
I vaguely remember I could not convince TOSLINK connection to HK AVR to transfer more than 96kHz. But my memory leaks a lot 🙂
of course not... most likely I will not use it for recording at all.
...but on the other end I'm quite intrigued by their ultra-low jitter clocks! (they claim jitter < 1ps !)
I very doubt that figure in the super-noisy PC environment. But no hard data to support my doubts.
Re: Re: Re: Re: Which sound card is best for Linux audio?
right... and of course it is so! I've been dumb stupid: this card does NOT have S/PDIF coax out on RCA, only Toslink optical! 🙄
(the two RCAs are for the 2 "main" channels analog audio outputs).
how does the guys at ESI have managed to get the Juli@ do it?
BTW, Audiotrak should be somehow related to ESI: they even had the "MAYA44 MKII" which looks a lot like the one with similar name from ESI... 😕
😀
BTW, I can see the picture... just one clock crystal is visible there. Too bad, otherwise it did look promising... 🙁
well, for sure being something non-standard, not every piece of hardware will support that. Perhaps only a few will. Did you manage to get it through via copper S/PDIF?
yup, me too... (that's why I specified that it is what they claim... 😉 ).
Nevertheless, I guess that if they made such an unbelievable claim, even if it is not really < 1ps at least it should be way better than any other card on the market. Maybe... hopefully.
right... and of course it is so! I've been dumb stupid: this card does NOT have S/PDIF coax out on RCA, only Toslink optical! 🙄
(the two RCAs are for the 2 "main" channels analog audio outputs).
how does the guys at ESI have managed to get the Juli@ do it?
BTW, Audiotrak should be somehow related to ESI: they even had the "MAYA44 MKII" which looks a lot like the one with similar name from ESI... 😕
unfortunately, Polish is not a language I understand...
😀BTW, I can see the picture... just one clock crystal is visible there. Too bad, otherwise it did look promising... 🙁
well, for sure being something non-standard, not every piece of hardware will support that. Perhaps only a few will. Did you manage to get it through via copper S/PDIF?
yup, me too... (that's why I specified that it is what they claim... 😉 ).
Nevertheless, I guess that if they made such an unbelievable claim, even if it is not really < 1ps at least it should be way better than any other card on the market. Maybe... hopefully.
Re: Re: Re: Re: Re: Which sound card is best for Linux audio?
The two crystals of Envy24 are hooked also to Juli's Xilinx which generates clock signals of the appropriate frequency, fed to Envy, switched permanently to external clock. This way the internal (buggy IMO) clock-generation circuits are avoided. The driver achieves the required fs not by setting envy's clock registers, as for other cards, but by setting a combination of envy's GPIOs hooked to the Xilinx. This results in two benefits:
1. The card can generate 176.4kHz SPDIF (unlike when operated in the classical mode by the old driver without the Xilinx support)
2. The card can detect fs of the incoming SPDIF stream. The AK4114/7 SPDIF receiver used in many envy24 cards requires a fixed-rate external clock source to be able to detect (by comparison) the incoming frequency correctly and let the driver read it. When envy is switched to the external clock mode, its crystals do not oscillate and the SPDIF receiver has no independent external clock source.
Whereas in Juli envy is switched to the external clock permanenty. Xilinx runs the oscillators, and if the card is in SPDIF-input clock mode, envy is timed by the clock recovered from the SPDIF stream, but the receiver is provided with the independent clock from the Xilinx to enable the incoming frequency detection.
As a side effect, since the driver knows the current fs, I think Juli is the only ice1724 card in alsa where DAC/ADC settings are modified for various incoming frequencies. Their analogue filters are fixed, thus they should internally oversample to a constant requency range for any input frequency (x2, x4, x8). For all cards, the driver sets appropriate codec values when frequency change is requested from userspace. In Juli's driver, the procedure is also called automatically when a change in the incoming rate is detected.
Google's translation tools provide some funny english 🙂
http://209.85.135.104/translate_c?h...1.html&usg=ALkJrhizy1XdtVuCSJfh6lgoBw8pk1o1gQ
The TOSLINK and coax SPDIF-out are fed from the same pin of envy (SPDIF transmitter output). The transformer winding of the coax output is buffered with several 74HCT... gates in parallel. The coax can output 192kHz, tested many times. I will test the Prodigy192's TOSLINK in digital loop and let you know.
I can't tell, but I think their price is in the incredible amount of channels they offer - writing optimized programs for their huge (and expensive) FPGAs is pretty expensive too.
UnixMan said:
The two crystals of Envy24 are hooked also to Juli's Xilinx which generates clock signals of the appropriate frequency, fed to Envy, switched permanently to external clock. This way the internal (buggy IMO) clock-generation circuits are avoided. The driver achieves the required fs not by setting envy's clock registers, as for other cards, but by setting a combination of envy's GPIOs hooked to the Xilinx. This results in two benefits:
1. The card can generate 176.4kHz SPDIF (unlike when operated in the classical mode by the old driver without the Xilinx support)
2. The card can detect fs of the incoming SPDIF stream. The AK4114/7 SPDIF receiver used in many envy24 cards requires a fixed-rate external clock source to be able to detect (by comparison) the incoming frequency correctly and let the driver read it. When envy is switched to the external clock mode, its crystals do not oscillate and the SPDIF receiver has no independent external clock source.
Whereas in Juli envy is switched to the external clock permanenty. Xilinx runs the oscillators, and if the card is in SPDIF-input clock mode, envy is timed by the clock recovered from the SPDIF stream, but the receiver is provided with the independent clock from the Xilinx to enable the incoming frequency detection.
As a side effect, since the driver knows the current fs, I think Juli is the only ice1724 card in alsa where DAC/ADC settings are modified for various incoming frequencies. Their analogue filters are fixed, thus they should internally oversample to a constant requency range for any input frequency (x2, x4, x8). For all cards, the driver sets appropriate codec values when frequency change is requested from userspace. In Juli's driver, the procedure is also called automatically when a change in the incoming rate is detected.
UnixMan said:unfortunately, Polish is not a language I understand...
Google's translation tools provide some funny english 🙂
http://209.85.135.104/translate_c?h...1.html&usg=ALkJrhizy1XdtVuCSJfh6lgoBw8pk1o1gQ
The TOSLINK and coax SPDIF-out are fed from the same pin of envy (SPDIF transmitter output). The transformer winding of the coax output is buffered with several 74HCT... gates in parallel. The coax can output 192kHz, tested many times. I will test the Prodigy192's TOSLINK in digital loop and let you know.
Nevertheless, I guess that if they made such an unbelievable claim, even if it is not really < 1ps at least it should be way better than any other card on the market. Maybe... hopefully.
I can't tell, but I think their price is in the incredible amount of channels they offer - writing optimized programs for their huge (and expensive) FPGAs is pretty expensive too.
- Status
- Not open for further replies.