Recording Multiple Inputs Over the 'Net?

TFGeditor asks: "Thanks to the advice of fellow readers from a previous Ask Slashdot, I now have a PC system optimally configured to produce professional on-air radio programs. Now I have a new problem: my radio co-host and I are in different cities located a few hundred miles apart. In order to give the show a real-time (i.e. 'live') sound, we need to somehow connect us so that we can produce a show complete with co-host banter, real-time interaction, and so on. I want it to sound as if we were both in the same studio. How can we do this? Will Skype or other VOIP applications do this without the result sounding 'tinny' (like a phone connection)? Are there other apps that will do a better job?"

Comments

POTS?

[oyenstikker]

Get a phone with an audio out, plug it into your soundboard/computer, and call him up.

Re:POTS?

[Spazmania]

Sure, because a 4khz bandpass filter sounds fantastic.

Re:POTS?

[swillden]

You do realize that many, even most, of the group conversations you hear on over-the-air radio are between people who are connected via POTS, right?

Yes, and they generally sound like crap. Yes, you can hear them, but they still sound like crap.

Not if you have a sound-warming tube amp, and gold-plated connectors.

ISDN

[Ubertech]

This may bust your budget, but there are many radio hosts at commercial radio stations who use ISDN lines back to the studio. The digital voice signal is good enough to make the remote broadcaster sound like they are in studio.

I'm sure there is a better, cheaper digital solution out there. Just make sure you have the bandwidth to handle it.

Re:ISDN

[denali99755]

I don't know what your system looks like, but if you have Pro Tools or any software that will run as a VST host, you can use Source-Connect to stream broadcast-quality audio from one of your systems to the other. Source Elements, the company that makes the software, claims that you need at least 300kbps down for it to work, although I would recommend going higher than that, personally.

The catches are that a. it costs $400 for the basic version (only allowing you a connection to one other user at a time), and you'll need an iLok, and b. there can be up to a second of latency. Due to the latter problem, I wouldn't recommend using it in a live broadcast setting, but you should be able to edit out the latency and keep it sounding natural if you're putting a podcast together for later distribution.

Hope this has been at least somewhat helpful--this is a fairly exciting new technology that's just making its way into commercial voiceover production, and once they iron out the latency issue (it is kind of annoying, especially for actors trying to do dialog) I could see it saving everyone a lot of money by doing away with ISDN entirely. And good riddance.

VOIP

[rlp]

I listen to a lot of podcasts on my daily commute. Most use some form of VOIP. Usually sounds fine (as long as they're not doing CPU or Net intensive tasks in addition to VOIP). Some of the podcasts do interviews with non-techy folks in which case they digitize an analog phone line or use VOIP through a gateway (Skype). For off-site interviews, podcasters use various types of digital voice recorders.

Two podcasters that have info about their podcasting technology on their sites are: Leo Laporte (http://www.twit.tv) and Glenn Reynolds (http:/www.instapundit.com).

Re:VOIP

[HTH NE1]

The only problem with VOIP is latency. It's a subtle thing, so whether or not it's a factor will depend on the type of discussion, but it can easily throw off comic

Ah, you can fix that in post.

timing

Depends how much you want to spend.

[Thumper_SVX]

If you are really serious about making it sound "professional", then you'll have to be "professional". This means (ideally) a dedicated link between the hosts.

I listen to This Week in Tech (twit.tv) every week and they encounter the exact situation you have. The way they deal with it is either with Skype (which sometimes causes breakup of one of the hosts due to lag or traffic), or they use an ISDN connection. The ISDN is the best "pro" solution because it allows good quality audio to be passed across a digital point-to-point connection. No lag, no problems. The only problem is that relatively speaking the ISDN is slow and expensive. However, if you want a reliable, lagless P2P connection there's really no better solution for the cost... your next option is a point-to-point frac T1 which can get really expensive. Of course, it depends on the amount of bandwidth you intend to use.

I do some part-time work in a recording studio where often a member of a band is "remote" (or in one case, none of them live in the same cities). Since we're talking multiple high-bandwidth streams the studio actually has several P2P T1's. The results can be awesome as we get real-time audio down the pipe at very high bit rates and resolutions... and the recording can be mixed in real time just as if the band members were there.

Body language might be a loss though. ISDN is good when you're pushing high-quality audio... but you won't be able to get video down that pipe as well. The best way I can think to deal with it is to use two connections; an ISDN for the audio and use an Internet connection with a webcam so you can each see the body language of the other. It'll isolate the traffic so that they're not tripping over one another, and the video feed seems to be the one you can most afford to lose (due to latency, lag, packet drops and so forth).

I wouldn't recommend trying to do a solution across the Internet unless you can live with an occasional dropout.

Also realize that if you're creating either terrestrial radio or podcasts, you have a certain amount of leniency since the quality is lower by default than HD Radio or Satellite. I'm all for spending what it takes... but there's no need to spend more than you need.

Finally, realize also that no matter what the final bitrate and quality of your finished product, the higher fidelity the original streams you mix together, the better. Higher bitrate and quality will give you "headroom" for compression.

Hi fidelity voiceIP

[TropicalCoder]

I developed an application that sends CD quality stereo audio over the internet in real time (one way connection). As input, it takes whatever audio is presented to the input of your sound card (which could be professional microphones, for example) and compresses it to 128 kb mp3 before sending via TCP or UDP packets. TCP requires at least 30% more bandwidth than UDP. For UDP, about 384 Kbits of bandwidth should do, while TCP may need up to 512 Kbits. In UDP mode, some UDP packets are returned to the sender to create a kind of handshaking to inform the sender that his packets are being received.

Audio is send four mp3 frames at a time, resulting in a latency of about one-tenth of a second for both send and receive. In UDP mode, there is the option of selecting some number of buffers so that the audio will be buffered to prevent drop out. Of course then lag will be multiplied by the number of buffers. On top of that you have the latency of your internet connection. Altogether, the lag could be quite acceptable if you have a good connection.

This application worked quite well in all my tests, but you could encounter issues with getting past a firewall or a DSL router/modem. Nothing in the software deals with these issues. I would be willing to "permanently lend" this application to you to experiment with, but you would need a certain level of technical knowledge to get past your router/modem/firewall. To use this application, you would mix the incoming signal from your partner with you own voice and music. Your partner of course will be monitoring the show. This software requires DirectX.

I developed this to teach myself about winsock. I don't know if there is any future in this software since it does not employ the RTP protocol for audio transmission and RTSP for audio signaling like a typical VOIP app, and it depends on mp3. However, it works very solidly and efficiently. I have thoroughly tested it both via the loopback on my computer, and over a computer network, with both TCP and UDP. I never managed to find someone capable of helping me test it over the internet. I would be happy to give a copy of this software to anyone wanting to experiment with it, and especially with anyone that has more than my minimal knowledge about resolving these issues like getting past the router/modem/firewall. If some other programmer thinks this may have a future and would like to colaborate with me on some project, that would be great. You will find my email address on my web site - just click on "Contact me" on the main page.

Think outside the box...

[tchuladdiass]

If you are only going for the live "sound", but aren't actually broadcasting it live, then you've got a simpler solution. Use whatever quality link you can put up with when talking to your co-host, but don't use that link's output in the final production. Instead, have your co-host also record his session from his end at a higher quality (with only his audio, not yours), and stitch the results together afterwards.

Re:Think outside the box...

[HTH NE1]

Well, still record the crappy audio. It will help to synchronize the separate tracks.

When I edited together a two-camera wedding shoot to DVD for a friend, the cameras didn't have the same timecode, and one of them had to change tapes frequently. I used their on-board audio to sync the images together, then another audio recording from the sound system to replace that (which had to be rate-adjusted due to it being just an audio cassette, so having the camera audio helped to establish sync).

If the cameras had their internal mikes disabled and recorded no audio, it would have been hell trying to get lip sync right. For a non-video podcast, you still want to keep the conversation's timing close to correct.

Professional?

[rueger]

"Thanks to the advice of fellow readers from a previous Ask Slashdot, I now have a PC system optimally configured to produce professional on-air radio programs" Hmmm... I remain skeptical, esp. when you're seeking advice from Slashdot. To your question, no, you're not going to use Skype or VOIP for a "professional" broadcast, for any of a dozen reasons. As noted, you need a Telos Zephyr or similar product. There are broadcast quality units designed to transfer audio back and forth over an IP connection, but Skype isn't it. Don't waste time here, check out a few radio trade magazines. And, uh, "professional" is much less about gear than about talent and proven broadcast skills.

Ventrilo VoIP

[WidescreenFreak]

I use Ventrilo every weekend with my nephew about 20 miles away and friend about 500 miles away during our network gaming nights. The sound is really good, it's completely "in conference" where anyone who knows the IP address could join in, and I've never heard the drop-offs or digital skipping that occurs frequently in Skype or Google Chat.

Apparently, Ventrilo also allows different sampling rates, so you might be able to pump through a higher bitrate to make the vocal quality better; however, I've never played with that function, so take that with a grain of salt. The default setting works well enough and doesn't sound like a telephone.

It's also available on several platforms. I run the server on my Sun Blade 100 with Solaris 9, but the three of us use the Windows clients for gaming.

Record seperately

[Dmala]

If you both have decent recording capabilities, the best way to sound like you're in the same studio would be to each record your own track. Talk to each other over the phone or VOIP or whatever using a headset, but also speak into a decent quality mic, recording locally. When you start, send a couple of blips over the phone and make sure it gets recorded on both systems, so you have a reference point to sync the files up later. When you're done, just have him send you his file. Load both files into an audio editor, line your blips up to sync them, and you should be good to go.

ISDN

[CokoBWare]

I know of a radio show in Austin, TX that is connected to the radio network located in MN through an ISDN line. It's clean, clear, and digital. I don't know the kind of equipment they use, but it is a direct digital channel between both points, and I would highly investigate this as an option. It may cost money, but it's likely worth it ($50-75/month my best guess). Check your local telecos.