 |
|
| update |
|
|
By nutela - Posted on January 5, 2008 - 12:53:21 (#22358)
Current version when comment was posted: Beta 2 |
|
|
Could you update the driver Korli?
|
|
| Mia MIDI |
|
|
By nutela - Posted on April 2, 2007 - 13:27:04 (#21515)
Current version when comment was posted: Beta 2 |
|
|
I'll have this card for my birthday from my lovely wife :-) Should be less than 120 Euro...
It should be supported with the newer versions of the driver or? (source Haiku cia, dev Korli)
|
|
| duh... |
|
|
By umccullough - Posted on December 24, 2006 - 00:24:47 (#21084)
Current version when comment was posted: Beta 2 |
|
|
So, you didn't actually visit the manufacturer's website?
They specifically state: "PCI based Professional Recording Systems"
I personally would have interpreted that as "more expensive than integrated audio"...
|
|
| holy crow |
|
|
By BeOS Mr X - Posted on December 24, 2006 - 00:19:27 (#21083)
Current version when comment was posted: Beta 2 |
|
|
CA$259.00 that is more of something a professional artist would need. Me on the other hand just want good quality playback with low noise ratio. Plus I have this odd audio crackle problem with my current sound device.
|
|
| Proof =P |
|
|
By Cyan - Posted on December 23, 2006 - 23:39:40 (#21082)
Current version when comment was posted: Beta 2 |
|
|
http://www.future-micro.ca/db/product.php?productid=308
From the top 3 sites of the first Google search I tried.
|
|
| then if it is so not obscure then prove it |
|
|
By BeOS Mr X - Posted on December 23, 2006 - 17:45:24 (#21079)
Current version when comment was posted: Beta 2 |
|
|
show me an e-tailer url that has the compatible device on their page for sale, has a picture of it, price and ships to canada.
|
|
| Not an obscure card! |
|
|
By Cyan - Posted on December 23, 2006 - 00:25:37 (#21077)
Current version when comment was posted: Beta 2 |
|
|
The Echo cards are pretty well-known as far as pro and semi-pro audio interfaces go; more respected than the likes of Creative Labs, etc.
For BeOS audio, it seems they represent the best possible solution at the moment. The sound quality is higher than even the better consumer cards, the signal routing is more flexible, and the price is pretty reasonable too; I'll be upgrading to one of these cards very soon myself.
A very big plus is the company who make the cards have released their driver sourcecode, and have a very responsive customer service department when it comes to user's technical questions. Try asking Creative Labs obscure questions about signal levels and output impedances some time. =P
I hope this doesn't sound too much like a marketing ploy; I don't work for Echo, nor do I own one of their cards (yet). However, I've looked around quite a bit for BeOS-supported soundcards, and these really do seem like the best solution right now -- no more expensive than a flashy, underperforming consumer card, but better quality and more flexible, with total native support.
Maybe the situation will change as Haiku's USB stack matures and USB audio drivers are finalized -- there's a lot of high quality external USB audio interfaces available cheaply now..
|
|
| Port one, or contact your vendor then... |
|
|
By umccullough - Posted on December 22, 2006 - 13:17:07 (#21076)
Current version when comment was posted: Beta 2 |
|
|
Port a driver then :)
You also have the power to contact your favorite vendor of choice and ask them to open the source for their driver, open the specs for their hardware, or write a driver for BeOS/Zeta/Haiku.
for a list of vendors that sell echo cards, you can look here:
http://www.echoaudio.com/Sales/Online/
|
|
| it is sad that such an obscure card gets the driver but not popular models |
|
|
By BeOS Mr X - Posted on December 22, 2006 - 12:57:06 (#21075)
Current version when comment was posted: Beta 2 |
|
|
http://www.ncix.com/search/?q=Echo&minprice=Min.+Price&maxprice=Max.+Price
none there !
|
|
| Hmm |
|
|
By nutela - Posted on October 20, 2006 - 08:35:40 (#20799)
Current version when comment was posted: Beta 2 |
|
|
Can't remember why I posted this crap here but to let you now the Ali chipset is not supported in R5 with the Sis 7018 driver. Seems that this driver uses old past R5 API's
Any news on the Echo driver btw?
|
|
| Aha! |
|
|
By nutela - Posted on March 20, 2006 - 06:06:17 (#19237)
Current version when comment was posted: Beta 2 |
|
|
I know what causes it, different cpu frequency. Since I installed Zeta on the laptop with the psu connected, the cpu freq was 1.6GHz on battery it's 600MHz. If I boot off battery I don't get sound at all and the media addon server hangs when shutting down or restarting media server. If I boot from PSU sound is fine. When I then disconnect the PSU the sound gets choppy.
|
|
| zeta's media_addon_server |
|
|
By nutela - Posted on March 19, 2006 - 12:55:18 (#19234)
Current version when comment was posted: Beta 2 |
|
|
Sucks : ( I don't get audio now but I did when I first installed it. Now it hangs during shutdown etc.
I guess the MultiAudio media addon is dependent on it? (How does it work?)
|
|
| partial fail (Gina) |
|
|
By Ga - Posted on February 7, 2006 - 11:23:25 (#19024)
Current version when comment was posted: Beta 2 |
|
|
Tested with Gina (Gina20, 20 Bit external, 24 Bit internal) under ZETA 1.1:
Sound comes, but only on the left speaker, and with alot rough background noise.
Regards.
Sebastian
|
|
| Cyan |
|
|
By korli - Posted on February 7, 2006 - 04:21:49 (#19020)
Current version when comment was posted: Beta 2 |
|
|
I posted a new version, allowing to change some settings without rebuild.
I also tested with echo3g with min sized buffers, and it seems 64 frames is the minimum as for emuxki. This could be an OS/Media kit limitation.
I also tested ok with 24 and 32 bits samples (playback and record). Haiku multi_audio and mixer media addons have been patched to enable this. R5 SoundRecorder display is weird but it records correctly.
Bypassing the mixer is a good idea to gain latency indeed. You can eventually bypass the media kit and talks directly to the driver (not sure you gain anything).
I forgot to try recording at smaller buffer sizes sorry.
|
|
| Latency tests, etc. |
|
|
By Cyan - Posted on January 20, 2006 - 15:15:14 (#18876)
Current version when comment was posted: Beta |
|
|
I've performed some crude measurements on the latency with the hacked emuxki drivers and ObjektSynth. Emuxki drivers were set to 2 buffers of 64 frames.
Latency was compared against a Roland JV-1080 (hardware MIDI synthesizer) by sending MIDI messages simultaneously to the computer's MIDI output (attached to the JV) and ObjektSynth; the latter was connected via the MIDI kit, no cables.
The envelopes were set as fast as possible on both ObjektSynth and the sound module, ObjektSynth was panned hard right and the JV hard left using an analogue mixer. The sounds chosen had plenty of high frequency content and sharp attacks to aid analysis.
The result was recorded onto analogue audio tape (adjacent tracks) and re-digitized later (see note 1). The resulting audio file was analyzed in a wave editor to measure the number of samples between the onset of the left channel and the onset of the right channel.
The delay was found to be in the region of 3 milliseconds.
With MIDI transmission latency at ~1 millisecond per "note on" message, this means ObjektSynth was around 4ms slower than the JV-1080. Other software synthesizers were tested briefly and were found to have a similar delay.
The JV-1080 is known to be a relatively slow-responding synthesizer, so the fact that the software synthesizer is 4ms slower might explain why it feels "laggy" to play.
However, I was surprised to find that the transition between "highly acceptable latency" and "intolerable latency" is so sharp; a matter of a few milliseconds.
Therefore, a reduction in latency of 4ms would be enough to bring it in line with the responsiveness of hardware. Bypassing the BeOS mixer may be the solution. Presumably a multi-channel soundcard would help a lot here, since there are "spare" outputs that the BeOS mixer isn't using?
I also tested a value of 32 frames for the emuxki drivers, but the audio output was incoherent and crackly. Does this mean that I would be unable to obtain 32 frame latency on the Echo cards too, or do they perform better in this regard?
Note 1: I've come across a problem with the hacked emuxki drivers; if I change the buffer size from the default 512 frames, recording no longer works correctly. A setting of 256 appears to drop half of the incoming buffers, making the audio sound "time compressed". 64 frames makes recording impossible, and crashes SoundRecorder. This is using R5, the latest binary multi audio addon, the beta Media Kit (Be) and the emuxki drivers compiled from source (~5th Jan 2006). Is there something else I need to change besides the define at the start of multi.h, or is it a SoundRecorder problem?
|
|
| latency |
|
|
By korli - Posted on January 15, 2006 - 09:43:32 (#18840)
Current version when comment was posted: Beta |
|
|
Cyan,
I assume the BeOS mixer adds its own latency, yes.
Any apps can theorically connect directly to the multiaudio addon, which exposes audio cards. The drawback is the app needs to support the audio format suggested by the multiaudio addon (which is the one reported by the audio driver). BTW recording behaves the same.
pmvn, one frame is a group of x audio samples, x being the number of channels. I didn't try such low settings yet, but I'll make these tests one day for sure.
|
|
| Frames & latency |
|
|
By nutela - Posted on January 15, 2006 - 09:10:55 (#18838)
Current version when comment was posted: Beta |
|
|
Korli, what's that equation ; 2*512/48=21.3333... ? What does one frame contain, one sample? or 1ms samples?
32 frames sound very cool, Cyan 2*64/48=2.6666... ms and 2*64/96=1.333... ms latency, that's *low*! But not enough for you?
|
|
| Korli - |
|
|
By Cyan - Posted on January 6, 2006 - 14:14:04 (#18774)
Current version when comment was posted: Beta |
|
|
Thanks for the response. I've hacked the Emuxki drivers to use a buffer size of 64 frames, and the latency is improved. However, it still feels a bit "laggy" to play; apps now report 5ms. I assume this high figure is due to delays in the BeOS mixer?
Anyway, I was meaning to ask -- what's the situation with multiple output support in the Echo drivers? Specifically, is there currently any way of sending the output of a (specially written) application to specific outputs on the card (e.g., 3+4) while the BeOS mixer output is sent to 1+2 ?
If so, is there any way for an app go "direct" to the extra outputs (bypassing the mixer?) in the hope of reducing latency?
Thanks again
- Cyan
|
|
| Cyan |
|
|
By korli - Posted on January 4, 2006 - 11:14:20 (#18747)
Current version when comment was posted: Beta |
|
|
Latency: currently hardcoded to 2 buffers of 512 frames, the resulting latency is 2*512/48 = 21.3 ms. It's meant to be updatable with the Media preferences in a "future" version. It is hackable easily in multi.h to values as low as 32 frames (emuxki driver is able to work with buffers of 64 frames).
Sampling rate: currently hardcoded to 48000Hz. It's meant to be updatable with the Media preferences in a "future" version. This is hackable (not easily though) in multi.cpp. 96kHz isn't tested yet.
Thanks for your interest.
|
|
| A couple of questions |
|
|
By Cyan - Posted on January 4, 2006 - 10:11:25 (#18745)
Current version when comment was posted: Beta |
|
|
This looks very good so far, though I don't (yet) have any Echo hardware to test it with.
I'm upgrading to an Echo card soon -- either one of the old Be-supported (20 bit) cards, or one of the newer cards; the decision depending on these drivers.
So, a couple of questions:
1). What is the latency of these drivers? My current SB Live reports 18ms in ToneProducer and ObjektSynth. Horrible to play softsynths live from a keyboard!
As I am developing a software sampler, my largest motivation for upgrading is obtaining 10ms latency or better, so it can be played in realtime.
2). Is there any way of changing the input/output sampling rate (perhaps by hacking it) to 44.1KHz instead of 48KHz?
The majority of audio is recorded at 44.1KHz, so running the soundcard at 48KHz guarantees very measurable aliasing during sample rate conversion (even with Dano).
Failing that, locking at 96KHz would be fine (both 44.1KHz and 48KHz material can be resampled to 96KHz with minimal aliasing).
Thanks!
- Cyan
|
|
| I'm thrilled |
|
|
By nutela - Posted on January 4, 2006 - 03:46:04 (#18742)
Current version when comment was posted: Beta |
|
|
Try Zeta or the Zeta Live CD. BTW this is really good news! Will 24/96 be supported?
|
|
| Cardbus Layla24... |
|
|
By delpy - Posted on January 3, 2006 - 07:26:15 (#18733)
Current version when comment was posted: Beta |
|
|
Unfortunately I don't have a BeOS installation on my laptop at the moment, but I've got both a cardbus Layla 24 and a vanilla Indigo so I'll try and sort something out to test this.
This is a very promising development (at least for me!) Thanks for the hard work.
|
|
|
|
|
