Does MCP-T do bit-perfect recording?

[FokeyJoe]

Hiya,

I've been trying to record from a dolby digital 5.1 source (set top box - seems to be doing 16-bit/48khz) with my onboard Nforce2 soundcard (it's an MCP-T (Realtek ALC658?) - mobo is an abit an7). It's been sort-of working. I can get a soundtrack captured that besplit can recognize and decode properly. But only VLC can play it back all the way through, and there are obvious glitches in the soundtrack - but it's mostly there. ac3fix didn't find a problem with it. But I can't seem to get it any better. I tried with an sblive platinum with the kx drivers, but that obviously isn't bit-perfect and keeps failing to decode properly with besplit. What I want to know is:
1) Is the MCP-T actually able to do bit-perfect digital recording? Are there any settings I should be aware of to ensure it is bypassing any encoders?
2) If a device isn't capable of bit-perfect recording, where does it go wrong? At the besplit stage like my sblive, or with the audible crud that I'm hearing from my MCP?
3) If the MCP can't do it, what is the cheapest optical in-based solution that I can get in the UK that will? I'm not averse to ebay!

TIA

[mikiem]

1st off I'll qualify this by saying I looked into ac3 transfer, couldn't get it working with my hardware, and dropped interest, so I'm in no way an expert... :P

THere were threads on this at both doom9 & driverheaven. From what I've read, the ac3 you record, when you can record it, is padded to sync with frames of video. It has to be modified with besweet to remove that padding.

The later high end creative cards (with breakout box etc.) might be able to record bit perfect -- different results reported, and creative makes so many versions of each card. KX drivers may, may not work, & same with software versions you'll find at driverheaven. The key task involved is bypassing the DAC entirely.

There were reports of various cards that could record the ac3 as broadcast, but many are no longer offered, some are fairly expensive (think m-audio has a few, & turtle beach had 1 or 2). But even attempting to capture the ac3 signal to my knowledge is not that common, so other then searching and posting in those 2 forums I haven't a clue about finding much more info.

[FokeyJoe]

Thanks for your reply. I believe I'm using the procedure you described, although I will double-check to see if ac3machine does any extra processing to the ac3 from besplit, if you are right about padding still being there. Though it sounds odd to be padding the soundtrack to the video, it's normally the reverse, otherwise you get breaks in the soundtrack. That's not what I'm hearing, but could be related, so I'll check it out.

As far as hardware goes, yes I'm aware the audigy's support bit-perfect because as the kx driver docs point out, they have a direct spdif record mode that bypasses the DSP. But before I dip my fingers in my wallet, I'll exhaust the possibilities with the very capable soundstorm first. Forums mention that it does some kind of passthrough, and that it's output is most likely bit-perfect. But nothing to say 100% certain that its' input is bit-perfect.

[mikiem]

"Though it sounds odd to be padding the soundtrack to the video, "

My *understanding* is that the video has a whole lot more data, & they want the frames of ac3 & video to match to maintain sync, so you've got this gap between where the audio frame ends and the video frame ends. When you record this stream your hardware knows nothing about what's going on -- just records everything -- so the gaps are included.

Ahhh... remembered something... From the "examples.txt" with BeSplitv0.82:

"6. Converting a Dolby-Digital-WAV into standard AC3 file
this will create the file track01.ac3 on e:\
BeSplit -core( -input live.wav -prefix e:\track -type ddwav -fix )"

"As far as hardware goes, yes I'm aware the audigy's support bit-perfect because as the kx driver docs point out, they have a direct spdif record mode that bypasses the DSP. "

Please take that with a bit of salt... I've tried it with the Audigy 2, one of the 1st batch, & it doesn't work. It's been a while since I checked, but the number of folks who've posted success were definitely in the vast minority. Since you don't need any quality to capture the ac3 stream, personally I'd buy the cheapest alternative that's backed by the manufacturer & user reports to work.

[FokeyJoe]

My *understanding* is that the video has a whole lot more data, & they want the frames of ac3 & video to match to maintain sync, so you've got this gap between where the audio frame ends and the video frame ends. When you record this stream your hardware knows nothing about what's going on -- just records everything -- so the gaps are included.

Oh, I see what you mean.. it's only getting the bursts of audio data as they appear over the time of parsing the transport stream.. yes, that makes sense.

BeSplit -core( -input live.wav -prefix e:\track -type ddwav -fix )

Doing That. Although, it makes no difference to VLC if this is done or not. With VLC you can just rename the wav to ac3 and it's okay with it. I also ran azid -N on the besplit'ed ac3 and it produced masses of E7 CRC errors, which isn't looking good.

Since you don't need any quality to capture the ac3 stream, personally I'd buy the cheapest alternative that's backed by the manufacturer & user reports to work.

From forum searches this tends to be the M-Audio cards.. but because I want optical spdif I would have to get the M-Audio CO2 as well.. which doesn't make it the cheapest solution. There might also be a spdif->usb converter, but it's coax again, but might be an okay price with the CO2 - I'm trying to track it down.

[mikiem]

RE: SPDIF convertors...

THere are several available, and even a few detailed plans that make building one seem trivial. You'll find links over at driverheaven. In US dollars I think they started at about the mid $20 range, and even Amazon had some decent pricing. Some included some sort of signal conditioning, making sure voltages up to spec etc. -- on the optical end I think there was some aplification as well, to go longer distances etc. SPDIF coax has advantages, like using co-ax to cross the room (optical can get expensive)

[FokeyJoe]

I have spotted one device and managed to find a corrobarative statement on its' abilities, the M-Audio Transit (bit-perfectness discussed here). And I can get it for about £45-50. It's got an optical in (I think it's a 1/8" jack TOSlink, but the unit possibly comes with a converter). Although I've also found a lot of posts from people who say it's a pain, though they are mostly posts in 2004, mac owners or haven't noticed the latency settings. There was even one guy who claimed that he managed to blow one up twice! But it sounded like he was trying to put a high gain device into its' PCM line in (a guitar). So, I could give that unit a try... see how it fairs.

But I still consider this a defeatist approach... why does it not work with the Realtek AC658? I've seen nothing to the contrary yet to suggest it can't do it.

[mikiem]

"why does it not work with the Realtek AC658?"

REALLY guessing here might take a look at the waveforms you're recording. I found somewhere that it's supposed to look like a steady stream of white noise -- all I ever got was a periodic (clock?) tick from the DAC, same as recording with no source, & that was zoomed in on the file all the way. Know you don't have a real wav file, but the method I tried from the doom9 discussion used audio software to record, so file was named .wav and it opened. Anyway, & it's likely a longshot I'm sure, might be able to see if your audio glitches are caused by loss of incoming signal, and that might suggest treating the signal before your PC got it.

"I've seen nothing to the contrary yet to suggest it can't do it. "

Don't know if that means anything or not... Info is so scarce I don't think there's a whole lot of folks who've ever tried, or would be inclined to try.

[FokeyJoe]

Describing the waveform is tricky... the best I can give is that it sounds like a fan with brushes on the end of the blades brushing against a snare drum quite fast. It's very loud, it mostly maxes the input sound. As a waveform, it looks like a hexagon with other noise intermixed with it.

Don't know if that means anything or not... Info is so scarce I don't think there's a whole lot of folks who've ever tried, or would be inclined to try.

That doesn't mean one shouldn't try. If we don't push the boundaries, we won't achieve anything. I hope that even if I don't solve it with my existing setup, that I will find some way to do it, and that hopefully others will find this post and learn how to do it themselves. Maybe enough interest might cause soundcard manufacturers to make a point of mentioning whether their input buses are bit-perfect or not, thus opening the market out to the mainstream for all to enjoy.

And I really shouldn't write posts after reading Danny Wallace books!!

[FokeyJoe]

Well I'm happy to say that I have at last got results... it's working! I bought an M-Audio Transit (which cost me £54 inc vat + P&P) and using iuVCR it's working. I've got the latency on very high, 2-channel 16-bit 48khz. I'm amazed that I can record the video right alongside it - I couldn't have asked for a better solution (well okay, maybe a slightly cheaper one ).

Of note, audacity didn't seem to work - it kept producing pops even when its' task priority was realtime.

I think it's inconclusive as to whether my onboard card could do it or not. I tested the same scenario with it and it didn't work. Maybe it needs better drivers (like some not written by nvidia!!).