Digital sound

[knuckles_mctug]

Hello everyone,

Hi, I been interested in audio electronics for a while and have only just come across this forum. I was wondering if there was a digital audio section. I'm interested in setting up a high bandwidth network to stream audio throughout my house and was wondering if anyone knew anything about it??
Thanks

Knuckles

 

[Sawzall]

Digital area for DIY?

While this is not directly on topic, this concept is gaining merit. Given the number of chips that are coming out - such as the ones that contain a DAC and AMP on one, which to me seems like a pretty cool project for some speaker builder - it seems that there will be a crossover from one area to another to another coming on. I know that I would like folks who might be playing with these to let us know what is happening. Secondly, with so many digital streams coming on, Internet now, sat, IBOC for radio, etc, this will just grow. Having heard with my own ears some of these (IBOC, SDAR, Internet) we are going to have to work to make these sound even reasonable.

 

[paulb]

Digital audio over wireless

I've been searching for some time for a convenient way to distribute high-quality audio all over my house, including the back yard. Knowing some people in the wireless industry has led me to think that the best solution is a digital wireless link of some kind (the remote speakers you see are, I believe, 900 MHz analog technology). So count me in as an interested party for a digital forum here.

 

[knuckles_mctug]

A few problems

Obviously the biggest problem trasmitting audio is the bandwidth required. PCM raw audio takes a massive digital bandwidth.
I'm not sure how to overcome this. Does anyone know about the S/PDIF protocol?? This is the usual link used between CD players and offboard DACs for example. This format transmits the required bandwidth for stereo audio, maybe the hardware could be adapted slightly to transmit over longer distances.
Any ideas??

Oh and by the way a digital audio forum would be a brilliant idea

 

[Sawzall]

Digital area for DIY?

I work in the broadcast distribution area - Satellite in particular, but when building IFL (inter facility links) we are using all kinds of solutions. Wireless, microwave, fiber, cable, ISDN, etc. All most all of this stuff has required multi thousand dollar equipment at each end, with codecs in the middle. All of that really high end trying to make the signal path as clean as possible (I work in public radio, so sound is actually important, as opposed to just transmitter saturation - no 110% modulation for us). Well out of the range of a DIY.

But the possibilites are opening up for us. 2.4ghz is open season in the US (up to one watt), especially since this area has been abandoned by the telcoms as they have moved up in gHz for more bandwidth.

Regardless of the method, what has become real clear is that we have to stay away from transcoding - changing from one codec, or bitrate or anything. We really need to distribute *one* thing. AES has sorta become the broadcast standard. MPEG4 may be the future, especially since it looks like the Asians and the German/Dutch have squared off to make the "super system".

So the system of the future may be distributing from source, all digital, all the way directly to the speaker. Perhaps even the individual driver, since it seems that it will be possible to design a DAC, "Crossover network", and Amp in a small enough package tailored and tuned for each driver. Video will tag along too.

The wrench in the system will be the intellectual property owners. For example, in the SDAR's (XM, Dogstar) there will be *no* digital output. Only analog - that is the deal they had to cut with the RIAA. Even with the actual digital signal being beaten to with an inch of it life by lossy compression. Same with IBOC (the method proposed for digital radio). Trying to re-digitize this will sound horrible. CD's coming out now are putting in data noise as ways to stop pure digital chains that might possibly lead to recording.

Whats that mean? We are going to have to hack it (I say that in the nice way.. all of us here are hackers in the traditional sense of the word. We like to play with our stuff, learn how it works, make it better, as opposed to just taking it the way the manufactures give it to us). Across the pond, ya'll have seemed to have beaten the Region coding on DVD's pretty quick. We are going to have to do the same on other formats soon.

Just some random thoughts....

 

[blmn]

Ethernet networks?

I don't exactly how or even if it's possible to apply this to High Fidelity, but I'm using IP multicast transmition to transmit audio and video to my corporation (26 floor building and other 9 branches) and and the results are very good. The people wich operate the system can use CD quality audio if they want. I think it is a cheaper solution.

 

[Sawzall]

Hi FI...

First thing is are we going to accept "is digital anything" Hifi. I think we are going to have to - so little "software" is analog. That being said, I think it is open season. In broadcast, we run AES on ethernet cable all the time. Kloss and Studer make pretty incredible mix desks that are basically computers now. People have been making records using ISDN links, with artists in multiple cities for several years.

I know that is holy war stuff, but if we have to have digital, at least we can work on making it sound decent.

Now, we just have to figure out how to do it cheap enough for mortals. I could design an ethernet soluton for your home using computer based parts, but that is a little too expensive. The broadcast industry is going that way - XM has 25terrabytes of spinning disks with hundreds of thousands of CD's stored on them. All available from a pick list - instantly. I want one of those.

 

[knuckles_mctug]

"They" have also designed a digital amp which uses a positive and negative cycle PWM of the data which then only requires a low pass filter. No analogue components in the whole chain!!
How would I set up an ISDN link around the house. I design electronics, embedded microcontrollers, (mostly 32bit cores), would an ISDN link be able to interface to a normal UART or SPI??
What would be good is if my main house PC can connect to a few microcontroller cards around the house which control the network and human interface.
Any help would be great.


[Edited by knuckles_mctug on 07-29-2001 at 04:12 AM]

 

[Sawzall]

Hmm..

I am speaking without actually designing anything here yet - so its a little out of my ***, as one might say - and it is 4:30am. Do we want wireless, or is wired OK?

Wired is easy. I have to look it up, but running it around your house is pretty easy with a computer. Commercial software routers for this are all over the place - bits is bits, the key is when to decode. Digigram makes a great card for this - up to four input/outputs at a time. Takes SPDIF and give you AES. They also make a AES/SPDIF transformer (See Cirrus Devices for some white papers), and once you get to AES, all kinds of equipment plays with it, mixers, computer programs, recorders, MIDI, even Genelec monitors play it out direct (and sound pretty damn good at it, too). You are then playing with the sound at the un-decoded bit level, so you will not be intoducing digital artifacts that are not already there. One issue here is that there needs to be a master clock in all of this so that all the timecoding signals are in sync.

Wireless is harder. You are now trying to do this and adding in a step. I suppose one could use the 803.11 (airport for mac folks)standard. These base stations and cards and such are getting cheaper, but still rely on ethernet. Problem with ethernet is that it does not garantee delivery on time in order of all packets. Obsticals to overcome. But in a house, under 300 meters, this should not be a problem at all. I know of radio stations that are using comsumer level equipment to make short bounces from studio to studios using the "rabbit" type TV equipment. The Video links have plenty of bandwidth and pass AES no problem.

A little work and we could come up with something. I know a commerical system that could do it in about a minute (it is just 4k a copy). Some of the less expensive stuff (ProTools, etc might with plug ins) Getting it cheap might be some work, but hell, thats half the fun in DIY. If you run a Mac, you are ahead of the PC crowd.

As I sit and write this, I see the convergence with home studio stuff. They are concerned with recording it, we are concerned with playing it. Funny how all this digital stuff is just getting into one big pile.

Have to quiz a buddy - he just put in one of these networks with over 80 workstations. I suspect he has this simple problem licked.

My fun is going to try to build my first true digital speaker. When it says digital ready on the front, it will mean it.

 

[Ignite]

Re: A few problems

Obviously the biggest problem trasmitting audio is the bandwidth required. PCM raw audio takes a massive digital bandwidth.

The answer to that is compression. MPEG-3, Ogg Vorbis, etc. Most of these technologies compress 10:1 with almost the same quality to the original data. The major problems with these is that the process to compress the audio with any precision takes much longer than realtime. LAME is about the most widespread, and definitely the fastest software mp3 encoder out there. But it is still far from realtime.

Software and hardware will eventually be able to do it in realtime, perhaps there is hardware that can now. However it's unlikely the RIAA wouldn't sue the pants off of anybody who tried, regardless of the use.

Online streaming MPEG "radio" is widely available (www.shoutcast.com for WinAMP, I'm sure Microsoft has reciprocated this idea with thier own Windows Media Player) nowadays, with many servers delivering 128kbit MP3, which is near CD-quality. It's only 16kbytes/s. A single DSL, cable or satellite link could listen to many servers at once at 128kbits. The answer is not in the airwaves, it's in the internet.

A sidenote: Windows Media Player format or RealPlayer format don't come close to the quality of a good VBR MP3. Ogg can make smaller files or slightly larger files with near-archive quality.

 

[hifiZen]

Bluetooth is another wireless possibility. if i recall correctly, bluetooth should support enough bandwidth to fit 16/44.1 audio uncompressed! (download only - upload bandwidth is smaller, probably for power consumption reasons, since bluetooth is intended for portable applications). I'd have to look at the specs again, but I think bluetooth may also provide an isosynchronous data pipe option, much like USB or IEEE1394 (firewire). And, 802.11 may still be a good option, since it's bandwidth well exceeds what is required for 16/44.1. All you really need to watch for in an ethernet environment is that you don't have much other traffic causing congestion and screwing up your QoS. With sufficient breathing room, IP can deliver real-time data very well.

 

[Mooze]

Audio bus

If real time is a requirement we'll never get there using compression.
The whole idea with compression is that you take a chunk and make it smaller. The compressor end needs to get that chunk, process it and send it, so when it reaches the receiver it's already not realtime - It doesn't care how fast you compress since you always have to get the chunk to have something to work on.
The smaller the chunks the closer to realtime, but real-realtime should be theoretically impossible I guess..

In this home-audio-bus application this should only be a problem when syncing with video, and probably the "chunks" can be small enough to make it work anyway. If not we'll need to delay the video to match.

As already stated, the bandwidth is the key.
Bluetooth is about 700kbit, so it should theoretically be able to carry five 128k streams. In real life it will be a lot less though, since there will be other BT devices around as well as good old interference.

How about a dedicated 10Mbit coax Ethernet?

</Mooze>

 

[hifiZen]

1394 / Firewire overcomes the real-time problem by using time codes and global clock syncronization through the interface. All this means is that you need to deliver the data slightly in advance so it can be prepared and buffered before it's scheduled playback time. 1394 includes so-called "isochronous" channels which allow for guaranteed delivery times and bandwidth, as well as priority over other network traffic...

Anyway, a really cheap way to go for DIY would be SPDIF/IEC958 over cat 5. Cat 5 has 4 twisted pairs, so you could pipe a full 8 channels of audio down one cable. The only thing is, since the characteristic impedence of cat5 isn't 75 ohms (I think its 120 or something around that value), then you need to make sure all your spdif receiver interfaces are properly terminated. The other alternative would be to include a small pulse transformer to match the impedences, but this also results in altered signal levels, so might not be a good way to go. Now, as far as distance and jitter levels are concerned, I really don't know. I suspect you could run several hundred feet without much trouble, but some form of jitter removal at each receiving unit would be a good idea.

 

[Won]

Mooze -

To state what you were suggesting more precisely, there is a tradeoff between compression ratios and processing latency, and this is dependent on "chunks" or block size (also, the number of blocks) that the algorithm uses to encoded the stream. This is an algorithmic latency; faster computers do not allow you to have less, only different algorithms.

However, it is certainly possible to have a compression algorithms (e.g. DPCM, MPEG Layer-1) that succeed at being "real-time," or within a particular latency-bound.

Just a few words from a software-ish guy.

-Won

 

[ajames]

Here is one simple idea:

I use a very simple process. All my music is stored as 196kps MP3s, stored on my server. 100Mps Cat5 network w/ switch.
Each station uses a 200MHz minimum PC with sound card. I can push it or pull it, depending on the results I want.

It's not elegant, but very efficient. Up to four stations can actually listen to different playlists without any diminished capability.

 

[R. McAnally]

I have heard good things about the Radio Shack 2.4GHz wireless video distribution system, working with SPDIF signals. I suppose you could buy multiple recievers and one transmitter, and transmit from wherever you want (eg CD player, computer, ect..).

Good luck

Hello everyone,

Hi, I been interested in audio electronics for a while and have only just come across this forum. I was wondering if there was a digital audio section. I'm interested in setting up a high bandwidth network to stream audio throughout my house and was wondering if anyone knew anything about it??
Thanks

Knuckles

 

[Rarkov]

I am trying to design a wireless 5.1 system for my final year degree. hifiZen (Chad) mentioned Bluetooth which I think is a fantastic idea. You can get 5 chips (albeit Ball Grid Array) for $70 which is about £45 for us Brits! The 10m range is perfect for a theater system but it may well require 10 chips to get 'real-time' performance.

The Active Xover will be done in the Preamp before entering the Bluetooth Transceivers and a small amp in the speaker will make it totally moveable.

As you will note, the projects aren't identical - but we may be able to work together to get a Mid to Hi-fi wireless system.

I also agree that any suitable digital coding would be able to recreate a fairly accurate analogue signal at t'other end so digital transmission is just about a must! PCM coding is used frequently for Voice with 8k samples per second...Sounds fast - but no where near 44k or the likes!

Gaz

 

[newmz]

a little expansion on ajames's idea;

it's a reasonably effective and cheap method to setup a machine (with ideally a reasonable processor and the biggest hard disk you can afford to store all the music) to serve audio as data (mp3's being the obvious example) to a few "bare-bones" client machines, especially if you have the know how to do it in linux - then the clients can be really low spec machines - you then don't really need much in the way of processing power, ram, etc because you can just leave out all the bells an whistles of a graphical operating system. they wouldn't even necessarily need a hard drive each and you can set them up to operate via just a keypad - no full keyboard or monitor necessary. then though, the "quality bottleneck" will be the soundcards that provide the audio line outs to whatever you're playing the music on.

as i was reading this i was wondering, what are the limitations of optical lines, in terms of distance, bandwidth, etc? you can get very high quality soundcards (with 24/96 converters, etc) that have optical and/or electrical digital in/out. BUT - how far can you run sp/dif or whatever without signal degradation?

 

[newmz]

after that last post i had a look around; "Cajun" is described somewhere as an app for turning your pc into a "massive audio jukebox". you can find it here.
it can send audio to multiple "client" amps or whatever if you have a multi output soundcard.

this is the blurb from the website:

<snip>

CAJUN is a project to turn an old computer you might have kicking around into the centerpiece of your car or home audio system. CAJUN's features include:

Multiple stream support. If you have multiple sound cards in your computer, you can connect one to bedroom speakers, and one to your living room speakers. Each stream is separately controllable; i.e., each has its own set of keypads, infra-red controls, and LCD displays. (This feature is not likely to be useful in a car, unless you have a really long car :^)
Each stream has multi-device support, including: multiple input devices and multiple output devices (displays). Both the front seats and passengers in your minivan can see what's playing on their own display. All displays are updated at the same time.
All Displays can be any size - no limitations with multiple streams.
Display content is completely customizable. Different-sized displays can display different data.
Input and output devices may be connected and disconnected at any time. CAJUN continues to run.
Supports keypads, infra-red remote controls, LCD/VFD displays, FM cards, CD players, MPEG audio files, (WAV files+Ogg/Vorbis coming soon), Shoutcast/Icecast streams, etc.
An extension for invoking arbitrary Perl/Unix commands.
Easy-to-use, documented Perl API for writing new drivers.

<unsnip>

 

[Yoda]

has anyone else used a 2.4ghz video distobution system for SP/DIF? this is interesting to me because its already done! cool!