PCM to PC to CD

[Linguini]

Hi,

I am a linguist and was hoping that HI-MD was the answer to my prayers. I needed a device that can record language in an uncompressed format. It appears that HI-MD will do this. However I have heard that to move the recording down from the HI-MD machine to the PC requires Sonicstage and that this puts some kinda curse on the recording meaning it can’t then be burnt to CD etc??? Sorry for the none techie speak but all I want to do is copy the uncompressed recordings to the PC and then edit them into individual tracks and burn them onto CD. Does anyone have experience with this?

Cheers.

[Linguini]

I think I have the answer - NO - after reading other posts on the subject. I will wait for the hacks or for Sony to come to its senses (no prizes for which comes first).

[dex Otaku]

I got my Hi-MD recorder last week and have been doing so every day since.

Look into Total Recorder. It is the current solution to these woes until Sony release the wave convertor.

[rirsa]

If you are doing professional work I wouldn't bother with MD. Get yourself a professional recorder. A Marantz PMD-670 doesn't cost that much more but it is packed with professional recording features you won't find on an MD recorder. And you'll be able to dump the contents directly to PC using a USB cable or CF card reader.

Lots of pro electronics stores sell it but Saul Mineroff Elecronics generally have one of the best prices form an authorized dealer.

[Capn002]

On the contrary. I am also a linguist and I would probably definitely go with MD. Maybe you already know this, but the whole "linguists must use uncompressed recordings only" thing is a bit of a lie.

First, just compare the physicality of the options. 3lbs for the Marantz, 11oz for a Sony DAT, 3 oz for a MD. Which do *you* want to lug through the rainforest?

Second is price. You could buy two or three MD recorders for the price of one Marantz or DATrecorder. Ridiculous.

Third, I can't say anything about the compactflash cards, but DATs are much much easier to destroy than a minidisc.

Finally, it depends on your purposes for recording. If you're recording just to make transcriptions, like for Conversation Analysis or similar, MD's are a beauty. Then, either keep the convos on MD and transcribe directly from that, or upload (see other threads for various discussion of this matter) and transcribe from computer.

If you're doing it for anything else except acoustic analysis, again, go MD. ATRAC uses psychoacoustic based compression. So, what's getting cut out (of language/speech), is what we can't hear anyway. It is information that is NOT PERCEPTUALLY SIGNIFICANT (or even *perceivable* at all (in theory)).

Now... if it's for acoustic analysis, then some real issues do arise. I've found as much information to support using ATRAC files for acoustic analysis as I've found against it. The simple truth is, linguists (phoneticians especially) mostly just assume that compression is bad. There aren't many actual studies that support this claim (one, in fact, and it's sketchy work), and I've found two studies that directly refute it. Simply, if you're analyzing vowel formants, durations, etc, MD is wonderful and even ATRAC compression is fine. We don't need PCM. Wm Labov even says so on his website. For transients or other highFQ stuff, ATRAC is not so good. DAT and CompactFlash recorders don't have this problem.....

But... the new HiMD's fix this. You can record PCM without any loss. Yipiee. And while you can't move it directly to the computer, blah blah blah. If you're a linguist, you're probably going to be listening to these recordings over and over again anyway, so real-time *uploading* is totally not a problem.

And really, MD's are so sexy. (And they double as a toy!) I've been using a MZ-N707 (recording in ATRAC even) for vowel analysis, and it works great. I did, however, succumb to the pressure for PCM and tonight I bought the new HiMD NH1. It should be here on Friday, and then I'll be doing acutal field work with it next week. I'll post reports when I get back.

Don't be silly... buy MD.

-d

[Christopher]

Welcome to the forums you two, and I really enjoyed your insight Capn - very good stuff. :happy:

[rirsa]

I didn't argue that linguists must use uncompressed recordings. I guess that depends on the linguist but that wasn't my reason for recommending the Marantz over the MD. I don't have an issue with compression. I'm not going to question the original poster's need. The PMD-670 will do PCM at various sample rates up to stereo 48KHz, as well as MP2 and MP3.

I do qualitative research. I record interviews and transcribe them. I've used several MD models, both Sony and Sharp. MD can work very well. It's not a bad choice. I just think that in most cases someone that is doing professional work will be better served by a machine like the Marantz.

On your other points: Weight and size. Yes the Marantz is larger and heavier, although you exaggerate a little. I also thought this would be a big issue for me before I got the Marantz, especially as I was used to using MD. In practice the machine doesn't seem that large and heavy. What I have found is that the design is extremely good and the increased size means there are no tiny little buttons that are difficult to use and the meter is huge and easy to see. When I was using a Sharp MD recorder I always took a check list of buttons I had to push to set the recorder up properly at the start of an interview. The Marantz remembers all the settings so it can be set up in advance. I get to the interview, put the Marantz in a suitable position, flick the power switch, flick the large red record button on the front and away I go. Very simple. It does the job with little effort so the user can concentrate on doing the interview.

What would I want in the rainforest? Probably the Marantz. The rainforest is an extreme environment. The Marantz has been out for over a year. Professional users (reporters and others) have given the machine some pretty heavy field abuse. I've seen reports of people dropping it and completely submerging it in watery mud, using it in extreme temperatures, having it sprayed with fire retardant while covering forest fires, and it keeps on ticking. Mine has been absolutely rock solid. I can't say that about my MD recorders. I always felt I had to handle them carefully. And on at least two occasions I had discs fail on me.

Price. Yes the Marantz is more expensive. I paid just under $600 for mine. You mentioned that you just bought a new HiMD NH1 which Minidisco lists for $380 so in your case the Marantz is only 1.5x more expensive not 2x or 3x. Your price comparison only works if you buy the absolute cheapest MD recorders available and it is still a little bit of an exaggeration. Of course there are other expenses and how you parse those is a significant factor. But also see comments on transcription costs below. For many applications the cost of the recorder, microphones, power, media, is all a bit of a red herring. In any case, my guess is that we'll see a smaller, lighter, simplified solid state recorder from Marantz in the next 12 months or so that will have a street price of $400. Ten years ago Sony was a big player in the cheaper end of the professional portable recording market. As a result of their various corporate anxieties I think they are abandoning that market to companies like Marantz.

Reliability of CF cards. Solid state. No moving parts. They are not indestructible and they do fail but there's no contest here (Try this with your MD media: http://news.bbc.co.uk/2/hi/technology/3939333.stm ). CF cards have been about for a while and are widely used in by amateur and professional photographers. They have a well established track record. I've never had a CF card in the Marantz fail. I can't say the same for MD media (not to mention the dreaded Toc error). I'm not sure anyone makes DAT machines anymore. DAT's time has passed.

Transcription. If you are doing a lot of transcription it is much easier and faster to do it using software (the software I use will sync up the audio with the transcription-you can't imagine how useful that is...). Also, if you are doing transcription, the cost of the recorder is trivial (which invalidates your price issue above). Transcription is time consuming and costly. On a single project we might budget $20K. You actually want to buy a recorder that will give you the best quality audio with a minimum of fuss. Saving a few dollars on this is a false economy. If you cut the transcription time down by just 5% you will rapidly recoup the cost of the recorder. I personally think the sound quality from MD is very good but the recordings I have from the Marantz are significantly better (even when using the internal mic). (Of course the quality in this case is partly a function of the way the recorder is used--what I'm saying is that for this type of application it is much easier to get a high quality recording using the Marantz.) While some demanding users do find the Marantz preamp wanting, you will get a much better preamp than anything you'll find in a consumer MD machine. In general, the electronics in in this machine are much better and it has a stack of very useful professional recording features you won't find on any consumer MD machine. You get good stuff for that that extra $220 (and the fact that a bunch of critical stuff doesn't have to be stuffed into a tiny case).

Transfer to computer. Some people don't have an issue with real time transfer; many people do (witness the endless moaning about Sony's position on this issue on these forums...). Anyway, just for the record, the Marantz has no copy protection. You can hook a USB cable to the recorder and the CF card appears as an external hard drive or you can eject the card and pop it into a USB2 or Firewire reader for even faster transfer.

"MD's are so sexy. (And they double as a toy!)". I won't argue with you on that one. The Marantz is a serious professional machine. It is designed to record interviews and meetings. It is targeted at professional users who need a serious tool that is well designed, reliable, easy to use and will provide excellent audio quality. It is also backed by a decent support. If your MD breaks, you've got to deal with Sony! Good luck.

In some respects this is like comparing apples and oranges. These machines are targeted at different types of user. Most people on these forums are probably best served by MD. If you are recording for a professional application, like those discussed above, the Marantz is probably a better choice. That said there is some overlap and there are some professional recording applications in which consumer MD might work better than the Marantz. For example, if extreme portability is a critical issue, go with the MD recorder. I suspect for most linguists the Marantz is a better choice. See http://www.ling.upenn.edu/phonetics/FieldR...dingAdvice.html .

Some user reviews:

http://www.rwonline.com/reference-room/pro...tz_PMD670.shtml

http://sldghamr.faithweb.com/PMD670.html

http://www.micsupply.com/pmd670review_htm.html

http://www.d-mpro.com/users/getdownload.as...?DownloadID=271

l

[Capn002]

Thanks kurisu! It's fun to be here!

[Christopher]

Rirsa, that's an excellent and well-summarized counterpoint that I see Capn has no objections to. I really value posts such as these as it puts a certain sense of maturity + class within these forums. I look forward to more posts from the both of you.

[Capn002]

the Marantz is larger and heavier, although you exaggerate a little.

Yes. The Marantz is actually 2.88lbs, not 3lbs. I was off by about 2oz.

Price. Yes the Marantz is more expensive. I paid just under $600 for mine. You mentioned that you just bought a new HiMD NH1 which Minidisco lists for $380 so in your case the Marantz is only 1.5x more expensive not 2x or 3x. Your price comparison only works if you buy the absolute cheapest MD recorders available and it is still a little bit of an exaggeration.

The price comparison works if you go by suggested retail price. For the NH1 that's $399, for the N707 that's $199, for the Marantz that's $899. You paid just under $600 for Marantz? great, way to price hunt. I paid $30 for my MZ-N707. I paid full price for the NH1 because I needed it quickly. Regardless, my original statement was not exaggeration.

And anyway, on a student's budget (or even a linguist's budget in general), even that extra $220 is a month's worth of groceries. It really does matter. Not to mention that a 1gig compactFlash card = $140 whereas a HiMD 1gig disc goes for about $10.

Mine has been absolutely rock solid. I can't say that about my MD recorders. I always felt I had to handle them carefully.

Marantz durability sounds awesome. It's funny, I was going to comment on MD durability over DAT. People have to treat their DATs like a crystal baby. Seriously, one of my profs wraps his in a diaper. Compared to DATs, MDs are like lead pipes. I just toss mine in a bag and go. I've been treating it with as much care as I treat any Walkman, and it delivers the same quality as when I first bought it. But seriously, if I dropped it, esp in muddy water, I'd cry and assume all was lost. Wow for marantz.

Reliability of CF cards. Solid state. No moving parts

Wow. That link was awesome. You've got me here. No contest. That's a good thing considering how expensive they are :rasp:

I'm not sure anyone makes DAT machines anymore. DAT's time has passed.  

Sadly, they do. They are *still* the standard in linguistics, even though they're big, bulky, you can never find tapes, etc etc etc. Actually, most of my post was supposed to be regarding DAT vs. MD. Because, as a linguist, those are about the only choices that linguini has. We linguists are a superstitious, tradition-based, pseudo-scientific lot. Therefore, since not many people have been using CF for fieldwork, not many people will, and those that do will open themselves up to having their work questioned because of it. This is what happened when MD first came out as an alternative to DAT or CC (tape). But enough people were enough satisfied that they're now commonplace. (but cf. my original post re:acoustic analysis).

If your MD breaks, you've got to deal with Sony! Good luck.

I've never had anything but good times with Sony Support. But then again, and maybe I should have said this at the very beginnig... I'm writing this msg on a VIAO computer, while listening to music over my Sony cdplayer, through my Sony receiver, while my Sony TV is on mute... and I have a Clie!

But for real, for linguists right now there are only two real options, DAT or MD. And the MD is still sniggerd at by phoneticians, so be prepared to defend yourself, though hopefully the HiMD PCM mode will quash this. Ladefoged (in Phonetic Data Analysis) actually doesn't even mention anything but DAT or recording directly to computer. Silly silly linguists.

-d

[Linguini]

Hi, Thanks for all the great responses. There are so many great points raised but not all specifically minidisc related. I think linguists are going through a painful period of readjustment with the new technologies arriving. HI-MD solves the problems inherent with MD but has MD (& HI-MD) had it’s day? With the new solid state/HDD machines now on the market, there is more choice than ever.

I think it is assumed that we all copy our recording to PC/Mac (BTW there are real issues with HI-MD and Mac which is unfortunate as I think there are a lot of linguists using Macs) and use software to edit the recording and burn the results. This can be done easily with solid state/HDD but is currently hobbled at the moment with HI-MD. Something that I think HI-MD has in its favour is the fact that the discs themselves are relatively cheap and can be kept as the source “original”. The problem I find with the other methods is it is up to you to ensure you make a copy of the “original” and keep it safely backed up (HI-MD enforces that). The analogy to this is photography and having a negative – how many digital photographers are going to lose a lot of memories with a HDD head crash. I really can’t imagine that anyone is really recording to a 1GB compact flash card and then putting it a drawer never to be deleted – no one has that much spare cash (150 bucks per 90 uncompressed minutes of recording?). There is mention of professionals using solid state and it certainly sounds like great equipment but I imagine these professionals have established work flows within organisations that ensure backups are taken and stored securely (possibly by a dedicated department). HDD recorders (iPod if it had a decent microphone adaptor) are okay but if you are in the field (up a Papuan Mountain or the Australian Desert) and your HDD goes kaput then you are probably stuffed. Of course this can also happen with MD but I you still have access to your already recorded media.

One major major positive in MD’s favour is battery life. I don’t know the specs but I do own an iPod and it dies after 4 hours away from the mains. I suspect the solid state machines eat up the batteries too. MD’s go on and on and if your battery is running low you can use the AA battery backup adaptor (and you can find AA batteries anywhere – even in a store in the Aussie outback).

I won’t go into the compression argument in linguistics. If you can do uncompressed recordings then I supposed you should. But the quite narrow frequency band that language is spoken would indicate that the advanced compression algorithms today would have a very little, if any, discernible impact on the recording.

So I really like MD (and HI-MD when it becomes un-hobbled) for recording language and I just wish Sony would realise that they have a product that could be something special if only they took the blinkers off and looked beyond the mass-market. How about a machine with the ease of use of the Marantz PMD-670K1 with HI-MD as the recording media (I must admit the little buttons and menus of the MD machines are a pain) now that is something that linguists would rush out to buy (and other recording professionals I’m sure).

[Capn002]

BTW there are real issues with HI-MD and Mac which is unfortunate as I think there are a lot of linguists using Macs

Oooh... excellent point. The soft lavender tones of my VIAO (the Mac of PCs) made me forget all about that. But, given that MD are still limited to real-time upload (for now), the mac problem goes away. I upload to Mac at school and VAIO at home (...strokes lavender keyboard lovingly). No problems.

I won’t go into the compression argument in linguistics.

Nor will I... we're a silly people. Here's the best argument I've found to counter those with compression phobia:

http://www.fon.hum.uva.nl/rob/Publications...DA2002paper.pdf But... since I'm not Ladefoged, I'm not allowed to make that call.

but has MD (& HI-MD) had it’s day?

When MD first came out, before Sony bought my soul, I lauged. "Who needs MD? We have our CDs, no one is going to go re-buy all their music... again. Sony is so lame!" That was, what? 1992 or something? Then, when Sony tried to push it as a *recordable* thing, rather than a CD replacement, I laughed. "Recordable CDs... that's the future, man. Give up Sony: you're so lame!" That was, what? 1997ish or something? Then, when I worked at a radio station, we switched over to MDs to play our commercials. I was like, "What? MD is still around? Why hasn't that died yet? We don't need no stinking MDs, we've got our carts! Sony is so lame!" And then I used them. They blew carts out of the water. This was like 2000. Then, I bought one. . .

. . .and the heavens opened up, and I saw the light, and I have been set free.

In short, no, I don't think they have any chance of dying out anytime soon. I mean, microcassettes are still around forchrissake. And as long as *line-in* exists, the current state of the art won't matter. PCM is PCM no matter what else might be available.

Am I rambling?

-d

[Capn002]

Hey moderators... would it be possible to move these to a "MD for Linguists" thread? I've seen a couple of us in various posts.

[aeriyn]

Hey moderators... would it be possible to move these to a "MD for Linguists" thread?  I've seen a couple of us in various posts.

Hmm... I think maybe you might have a good idea there.

Since MiniDisc seems to be widely used by linguists (a use I'd never even thought of myself, since I'm just a geeky music girl) and... well, you guys have posted one of the most intelligent threads in the entire forums.

A linguist sub-forum might be a nifty idea.

[dex Otaku]

Hey. If a linguist sub-forum might be a nifty idea - how about a field recordist / environmental recordist sub-forum?

[rirsa]

I'm a social scientist! How about a forum for users who are involved in tasks that involve the recording of human speech for analysis or some other purpose.

There are also groups of users who record nature and environmental sounds but there are some well established and active forums elsewhere on the net for those users.

[rirsa]

Linguini, Capn002, Kurisu and others:

Happy to contribute something here. Over the years I've learned a lot from the Minidisc forums, both the new ones and the earlier ones.

Most places sell the Marantz for around $700. You have to look around to find a price around $600 from an authorized dealer. I think you have a valid point that not everyone has this money. And flash cards are still relatively expensive although the rise of digital photography has the price halving every 12 months or so. When I was a graduate student I certainly didn't have this type of money to spend. MD is a good option that is more affordable. There still isn't much serious competition in the sub-$500 market.

The Marantz is a breakthrough product in terms of cost but I don't see why one shouldn't be able to buy a decent solid state recorder at a cheaper price point. I'd still be a minidisc fan if it weren't for the ridiculous upload issue. Sony is Sony's own worse enemy. The Marantz is overkill for much of what I do--it features I don't use or need. I think there is a diverse market of people that have fairly straight forward recording needs (e.g. in government, the legal profession, medicine, academia, business, etc.) but want decent audio, reliability, easy of use, and need to rapidly upload to computer without any ridiculous and cumbersome restrictions and without breaking the bank. The earlier models Marantz sold cost over $1000. Most other pro recorders (whether based on solid state, MD, hard drives, etc.) still cost over $1000 and sometimes much more. DAT seems to be on the way out. I'm not very impressed by other cheap options. Some people get excited about jukeboxes (iRiver, Creative Nomad, etc.) but I think recording in those devices is something that engineers added on after the fact or as part of a crude "everything plus the kitchen sink" approach. Recording functions don't seem to be very well implemented and my guess is that the reliability is probably less than stellar. I think a good recorder really needs to come from a company that understands recording and is dedicated to building a device in which recording is a primary function. There's not much in the middle range between cheap portable recorders that aren't that good and high-end pro audio machines.

Social scientists are probably worse than linguists when it comes to recording technology (and technique). Many appear to be quite happy recording interviews using cheap $50 tape recorders with internal mics that pick up the sound of the transport mechanism and which they insist on placing a less than optimal distance away from the speaker. It is a mystery to me why someone would put so much energy into a task to have their data recorded by a device that is difficult to use effectively and almost guaranteed to destroy at least 20% of their data and make the rest an unpleasant listening experience.

Archiving. Yes, I think an advantage of MD is that you have a cheap original media that provides pretty good storage. I'm not sure how long data will last on an MD but I imagine it fairs pretty well if not very well compared to other media. No one is going to store stuff on flash media--you get the data and you offload it. CD and DVD media is cheap but I think one has to be careful to use good quality media (e.g. http://www.mam-a.com/products/gold/archive.html). For really long term storage another issue is how long the players for any of these media will be supported. At some point the data will have to be copied.

A counterpoint to my earlier counterpoint on what to take to a rainforest: MD may have a storage advantage if one plans to be in a remote location for an extended period of time. If you can't offload the CF card onto a harddrive and create backups onto CD and DVD media then MD has a certain advantage. Lots of devices have been developed for digital photographers to offload flash memory cards though but that's another extra cost. A stack of 20, 50, 100 or however many MDs one needs may be a cheaper and safer way to go.

One shouldn't overlook the professional MD recorders. Marantz used to make an MD model but I think that was recently discontinued. I don't know if they plan to come out with a Hi-MD recorder. HHB makes a pro minidisc recorder called the Portadisc. It costs and arm and a leg and I've heard that it eats batteries but it has a very good reputation and is extremely durable. Again it is unclear if HHB will develop a Hi-MD version. BTW the current portadisc is advertized as the "First and only MD portable to feature a USB (Universal Serial Bus) interface for transferring audio to and from Windows-based PCs and Macintosh computers." If I remember correctly, this still happens in real time but there is no weird Sony software and DRM nonsense.

Battery life. I'm not into electronic gadgets that use weird batteries either. If your batteries die you want to be able to get a replacement that you can buy cheaply and find practically anywhere. A nice thing about many MD recorders and the PMD670 is that they will use Alkaline or NiMH AAs. The battery life on the PMD-670 is fairly good if you the latest rechargeable NiMH batteries (e.g. something rated 2300 mAh). I'm using 2100 mAh and I've never had a problem. I've seen people claim they can get 10 hours record time if they don't use CF microdrives and don't use phantom power for the mics, and I believe it. However, the PMD-670 does use eight AAs versus one in my Sharp SR-60. I'm curious about the battery life of the new Hi-MD recorders. The reports I saw were that if you were recording in PCM you weren't going to get anything like the amazing record times of the earlier generation of MD recorders although you'd still get respectable record times.

[The Prince]

This may be a dumb question, but does Hi-MD recorders in fact have the ability to record in PCM format? Is PCM compressed? I always thought it wasn't.

and

What format do the linguists use? What's higher?

thanks

[Ppk3000]

does Hi-MD recorders in fact have the ability to record in PCM format?

Yes.

No. (Not the native raw PCM form Hi-MD uses)

[dex Otaku]

No. (Not the native raw PCM form Hi-MD uses)

"Native raw form"? There is no other form of PCM. Period.

PCM = Pulse Code Modulation, often called Linear PCM or LPCM.

The whole point is that it's all the bits, all the time.

[Sbetsho]

Sorry, but could someone explain what does it mean to record language to MD? If it is just speech wouldn't 48kbps be just fine?

[Capn002]

What format do the linguists use?

That's one of our biggest problems, IMO. We don't have a standard per se. Some journals and conferences require PCM. And everyone always complains when someone uses compressed data, but major major research has been published based off of speech recorded over telephone lines.

What's higher?

I've been wondering the same thing, kinda. I mean, PCM is raw data, all the bits that go into the machine come out of the machine. But then there's 44.1kHz vs. 48kHz on DAT recorders (and computers add 22kHz and on down), and there's 16 vs. 8 bit settings (or something to that effect).

Does anyone know how the quality of low-bit PCM compares to high-bit compressed data? Like 22.5kHz, 8-bit PCM vs. 44.1kHz, 16bit ATRAC. How would those compare? In terms of both listening quality and amount of data captured.

-d

[Latexxx]

The quality of ATRAC(3(plus)) depends heavily on bitrate. 256 kbps ATRAC3plus (44.1 kHz, 16 bit) does definately beat 22.5 kHz, 8 bit PCM on speech (and music) content. On the other hand, ATRAC3plus at 64 kbps is ok for speech and offers 34 hours of recording time on one Hi-MD disc.

[dex Otaku]

Does anyone know how the quality of low-bit PCM compares to high-bit compressed data?  Like 22.5kHz, 8-bit PCM vs. 44.1kHz, 16bit ATRAC.  How would those compare?  In terms of both listening quality and amount of data captured.

Amount of data captured = [sampling rate * bit depth * no. of channels] bits per second; take that and divide by 8 = bytes/sec; take that and divide by 1024 = kilobytes/sec.

For example, CD audio uses a 44,100Hz sampling rate at 16 bits with 2 channels, which = 1,411,200 bps = 176,400 Bps = 172.27kBps.

Sampling rate primarily affects frequency bandwidth. In plain English, the higher the sampling rate, the higher the frequency that can be recorded, keeping in mind the Nyquist frequency - that f/2 is the highest frequency that can be recorded with any accuracy, where f is the sampling frequency.

I.E. CDs sampling frequency of 44.1kHz means a Nyquist freq. of 22,050Hz. A rolloff [equalisation] is applied before that freq. to ensure that the distortion that occurs near and above the Nyquist freq. can be avoided. [For those who don't believe me, try putting frequencies up to double to sampling rate into an unfiltered A/D converter and see what the resulting distortion is. It's pretty cool, actually.]

Anyway, the higher the sampling rate, the higher the recordable frequency, hence the higher the perceived quality.

Bit depth [quantisation] affects dynamic range. The general rule here is that you get about 6dB more dynamic range for every bit of resolution. 8-bit quantisation gives you roughly 48dB dynamic range; 16 bit gives you 96dB; 24 bit gives you 144dB [which exceeds the range of human hearing].

There's more to it than that, of course. Linear PCM works by slicing an analogue signal [alternating voltages that generally range from 20-20,000Hz and for the sake of argument, assume a line level signal of, say, 1 volt peak-to-peak, i.e. top to bottom] into a time scale [that's the sampling rate] and an amplitude scale [that's quantisation, or bits].

If you use 8 bit quantisation, you get a total of 255 slices from top to bottom [with one bit to say positive or negative].

With 16 bit quantisation, you have 65,535 slices from top to bottom. Can you see how this might make a slight differecne as to how accurate the signal that comes back out is?

And 24 bit.. that's 16,777,216 slices.

It's all about accuracy.

Many lossy compression schemes sort of work independently of bit depth, though. Because of the various methods of analysis they use to split the signal up from the time domain into the frequency domain, the number of bits coming in becomes sort of irrelevant as long as they're above a certain level [say, 16 bits just for arguments' sake]. Compressing 24-bit audio directly to atrac doesn't necessarily give a more accurate result than doing so from 16-bit data. And decompression into 24-bit rather than 16-bit does not mean a higher resolution signal on output, either. Garbage in, garbage out.

I'm way off topic now.

[rirsa]

Just to add a few points to dex Otaku's very detailed post:

Normal speech frequencies are in the mid-range i.e 250 Hz to 8kHz. You don't necessarily want low frequencies or high frequencies but it depends what you are doing. If you are recording speech for transcription the low or high frequncy stuff is just noise. If you are recording an interview for radio transmission those other frequncies might be ambience or atmosphere.

Given the Nyquist issue, (see previous post) that means one should use a sampling frequency of at least 16kHz and probably 22.05kHz. I use the latter sampling frequency and it works fine. If you use Cooledit you'll see that Cooledit has a MP3 preset for voice recording: 22.05kHz sample rate, mono, 16 bit (ie a 32kbps MP3) with a frequency bandwidth limit of 8kHz. If you are a linguist and you want to avoid compression try using 22.05KHz, 16 bit PCM.

I think most telephone systems limit frequency transmission to between 400Hz and 3.4kHz--the absolute most critical speech frequencies. So if you are recording a phone conversation I guess 8kHz might be adequate. (NB in most places it is illegal to do this without consent.) Phones aren't exactly hi-fi devices. They are engineered to maximize speech intelligibility while conserving transmission bandwidth (there are various codecs that are used for this as well...).

Practically all digital recorders use at least 16 bit sampling. I would avoid 8 bit.

Here's a link I found yesterday that provides a much more detailed discussion of these basics and other matters:

http://www.ling.upenn.edu/courses/ling538/.../AudioData.html

[Capn002]

Thanks for the info...

256 kbps ATRAC3plus (44.1 kHz, 16 bit) does definately beat 22.5 kHz, 8 bit PCM on speech (and music) content.

I'm totally not doubting you, but do you have any sources that would back this up? I'm required to use some citation other than bbs.

-d

[Capn002]

thanks everybody for the info... especially the stuff explaining PCM, which I haven't done any reading on. Linguistics books take it for granted without ever explaining it. But I guess what I'm really wondering is: Is there some kind of simple slide rule that would equate compressed and uncompressed qualities. Like a java-script page or something? 8bit, 11.5kHz PCM = X ATRAC = Zmp3, etc etc.

-d

[Linguini]

Thanks capn002, dex, rirsa and all – the information presented is very thought stimulating indeed. I am going have to reread dex’s post a couple more times to get it to sink in. This is off the subject of HI-MD now, but the microphone I use has a frequency response with an upper limit of 18kHz. I assume this means, taking into account the Nyquist issue, if you want to record all frequencies captured by the microphone you need a recording device that has a frequency response greater than 36kHz.

[Topaz]

Excellent discussion so far, everybody (and well moderated too). I'm ALSO a linguist needing to record lectures, conversation, interviews and other forms of human speech. I have my newly-purchased MZ-NH1, which I pre-ordered from my local Sony Shop here in Hong Kong, and while there are some annoying built-in restrictions to its possible operation (which I am none too happy about), it still serves my purposes very well.

I agree that different sub-forums should be set up with the "Hi-MD applications" as the theme - linguistic and language-related applications would be great, with further subdivisions of studio work and fieldwork, etc. I also think language learning, transcription techniques, and plenty of other possible topics could really benefit from the input of the forum users here.

Thanks for the lively and informative discussion. :grin:

PS: Still no sign here of spare LIP-4WM batteries and 1GB discs - I was patient enough for the NH1, I guess I'll just have to wait a little longer. :wink:

[dex Otaku]

..I guess what I'm really wondering is:  Is there some kind of simple slide rule that would equate compressed and uncompressed qualities.  Like a java-script page or something?  8bit, 11.5kHz PCM = X ATRAC = Zmp3, etc etc.

No.. not really.

PCM is PCM .. Sampling rate and bit depth determine the accuracy, period.

Any lossy compression scheme is dependent on what algorithm is used to encode the signal. Since that encoding is not linear, and does throw out part of the original signal [lossy = data reduction] there's no concrete way to quantify except to measure the difference between what goes in and what comes out - i.e. to either measure it somehow [which is difficult to do since a lot of compression schemes are not frame-accurate like PCM is - note that ATRAC -is- frame accurate because it uses fixed frame sizes and data rates] or to simply compare by listening.

Generally speaking, the lower the bitrate, the lower the quality - though as compression algorithms become more advanced, lower bitrates do increase in quality. As an example, try comparing atrac3plus [c.2003?] @48kbps with full-bandwidth mp3 [c.1988 or so?] @48kbps. The difference is staggering.

What it ends up working out to is that the less you throw out [i.e. the lower the data reduction, or compression], the smaller the differences will be between what goes into the codec and what comes out of it.

Another point is that comparing PCM at different sampling rates and bit depths with compressed sound [through any codec] at various bitrates is basically like comparing apples with automobiles.

The differences between various rates of PCM are basically obvious to anyone who has a good grasp on how [digital] audio works; half the sampling rate means half the frequency bandwidth; half the bit-depth means less dynamic range. It's very straightforward.

Psychoacoustically compressed sound, on the other hand, depends on many more things, such as: the frequency bandwidth of the material to be compressed; what kind of analysis is being used for compression [DCTs, FFT, wavelet, etc.]; how many bands and sub-bands are used for encoding [since this is what actually determines the resolution]; whether bandwidth-limiting is used, and whether it is variable or fixed; etc.

Whereas the differences, the artifacts of various PCM rates are fairly evident just by looking at specs - it's pretty easy to tell what to expect, really - the artifacts, and the severity of those artifacts, of every psychoacoustic compression method are essentially different from all others.

The only way to tell is to compare them and find out, whether by some means of technical measurement or by listening tests.

Most of the information you find out there will be subjective, of course. Measurements simply can not tell the whole story every time - so human judgement comes into the equation. At that point, you're basically having to decide whether to trust the opinion of someone else or not. Or - to simply go and make the comparisons yourself, should you have the means to do so.

[aeriyn]

PS: Still no sign here of spare LIP-4WM batteries and 1GB discs - I was patient enough for the NH1, I guess I'll just have to wait a little longer. :wink:

You can order the LIP-4WM from Audiocubes. :happy:

[rirsa]

Several other comments on compression that may be of interest to linguists and others:

1. ATRAC vs. PCM. Over the years this has been a hot topic for researchers and enthusiasts who record nature sounds (birds, etc.). Some people have been fairly negative about ATRAC (e.g. http://www.birds.cornell.edu/MacaulayLibra...ntOverview.html ); others think the critics of ATRAC have no idea what they are talking about (e.g. http://groups.yahoo.com/group/naturerecord...ts/message/5772 ).

2. Looking at some of the information on linguistics web pages I didn't see much about other issues but maybe I missed discussion of other technical issues. Sampling, bit depth and compression are just one set of issues. A high sample rate and PCM by itself guarantees nothing. You need to consider the mics, the cables, the connectors, the preamps, how the mics are used and so on. There is a whole system to be taken into consideration. Other factors may be much more significant than the compression issue in shaping the final representation of whatever it is you are recording..

3. I haven't seen any mention of level control. If you are going to be really fussy about artifacts, I'd worry much more about whether the recording was made using AGC or not. It is convenient to use AGC for many types of recording but for many uses it should be avoided like the plague because:

a) AGC compresses the dynamics.

The AGC circuits always involve some type of delay so they are poor at handling rapid changes in sound volume.

c) The AGC circuits adjust to anything the mic picks up, regardless of whether it is the sound you are focusing on recording or a background noise.

4. One advantage of PCM is if there is any type of editing. You don't want to be editing and saving and reediting and saving over and over again using lossy formats.

5. The bottom line is that any recording is a representation. Different types of system are appropriate for different types of uses. Some users will inevitably be more demanding than others but no matter what you use and how you use it, you are still using a tool to create a fixed representation of an event and that representation will never perfectly capture that event.

6. Researchers, particularly those involved in ethnomethodology and conversation analysis have written about some of these issues. Here are a couple of references:

* Modaff, J. V., and Modaff, D. P. (2000) Technical notes on audio recording. Research on Language and Social Interaction. 33 (1), 101-118.

* Ashmore, Malcolm and Reed, Darren (2000) Innocence and Nostalgia in Conversation Analysis: The Dynamic Relations of Tape and Transcript. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research [Online Journal], 1(3).

[rirsa]

BTW Anyone looking for a very nice open source audio editor should check out Audacity.

See http://audacity.sourceforge.net/

[sup8]

So Audacity works for recording through the usb? i have Audacity and just recieved my hi-md800 yesterday. going to test it tonight. do i even need to install sonic stage software?

[jadeclaw]

You have to install SonicStage, otherwise the USB playback won't work. And it makes titling so much easier. :smile:

Select Wave as the recording source in the mixer panel and set the slider to 100%

Start SonicStage, insert the MD and when the contents appear, start recording in Audacity, then start playback in SonicStage.

Only recordings made in Hi-MD mode can be played back via USB. Old recordings must go via LineIn of the Soundcard.

[Capn002]

rirsa,

Thanks for the refs... that's awesome... I'm doing a full test on the MD sometime this weekend. I'll post anything interesting.

-d

[Linguini]

Capn002 really looking forward to hear how your testing goes.

[jonnybiggs]

I have a question, a serious question. Would anybody like a bagel?

No seriously, I think I have grasped (sort of) the PCM/Hi-SP compression comparison, so can anyone tell me (if I didn't blip over it in all that scientific talk) is Hi-MD PCM "22.5 kHz, 8 bit" as quoted by Latex?

And then how does that effect PCM as a mastering medium? Is Hi-SP then better?

I'm not a linguist, rather a DJ who likes the idea of mastering mixes straight from my mixing desk and being able to fiddle with them (ie track marks) before burning a CD.

I guess you'll know I'm new around these parts, so apologies if I'm in the wrong forum...

[dex Otaku]

so can anyone tell me (if I didn't blip over it in all that scientific talk) is Hi-MD PCM "22.5 kHz, 8 bit" as quoted by Latex?

And then how does that effect PCM as a mastering medium? Is Hi-SP then better?

Hi-MD PCM is 44.1kHz, 16-bit. Which is the same standard that has been the basis for CDs since their development.* Which is the same standard used by most digital master tapes from U-Matic in the 1970s to DAT and beyond.*

Hi-SP is data-reduced from a 44.1kHz, 16-bit source. [for now - assuming 24-bit A/D-D/A convertors get put in there at some point.]

If you take -away- from PCM to get Hi-SP, how could it possibly be better?

* - this is not counting the days when many masters were actually 14-bit quantised but written in 16-bit words to the masters. I have at least 2 14-bit mastered CDs still kicking around here from the very early 1980s - but the standard was still for 16-bit. They just didn't have 16-bit A/D convertors everywhere at that point. Or something.

[Latexxx]

Hi-MD PCM is 44.1kHz, 16-bit.  Which is the same standard that has been the basis for CDs since their development.*  Which is the same standard used by most digital master tapes from U-Matic in the 1970s to DAT and beyond.*

Dat is 48 kHz, isn't it?