UTOSC: Inexpensive audio and video recording

Clint Savage set out to inexpensively record audio from Linux user group (LUG) meetings that could be easily shared. Soon after he started making podcasts available, though, he started getting requests for video. That led him to investigate cheap video hardware as well as free software solutions for video. At the Utah Open Source Conference (UTOSC), Savage described his journey and the solutions he has found.

Near the beginning of his talk, Savage stressed that he is not an audio/video engineer, just an interested hobbyist, and that he was presenting tools and equipment that worked for him—not necessarily the "best" way to handle audio and video on Linux. He started out very simply, using a $7 microphone from Radio Shack to record LUG meetings. He used Audacity to add in some introductory audio and then made the audio available as podcasts.

Once he had that working, he investigated live streaming with a Shoutcast-compatible server. He tried Icecast, but it "took a long time to set up". In the meantime, his listeners wanted video, and he wanted to stay with free and open solutions while not breaking the bank, so he started looking into cheap video hardware.

For cameras, he settled on a Canon ZR80 MiniDV digital camcorder, but said you can also use almost any webcam, including those built-in to laptops. He wanted to be able to do screen capture for presentation slides and screencasts, which led him to the vga2usb device from Epiphan. While that works well, it costs around $300, so he found a much cheaper solution using a scan converter and EZCap, which does a reasonable job for around $60. Lowering the cost allowed him to outfit each of the half-dozen UTOSC presentation rooms with systems capable of recording audio synced with each talk's slides.

For video capture software he looked, briefly, at the proprietary offerings for Windows and Mac OS X, but was quickly put off by the price. Not only is there the "Apple tax" on hardware, which makes it more expensive than the equivalent Windows or Linux box, but then you need to spend something like $3000 for video software. Windows has similar offerings, but it is just as expensive.

For Linux, there are multiple choices for video editing (Kdenlive, Kino, PiTiVi) and for screen capture (Istanbul, gtk-recordMyDesktop), but none of those directly provide streaming solutions. Flumotion is a free streaming server, but he spent three days and "never got it working". Most of those tools use GStreamer under the hood, though, so he started looking into that.

GStreamer is a library and set of command-line tools for handling multimedia content. Savage showed several examples of using GStreamer from the command line, starting with a very simple:

    $ gst-launch videotestsrc ! ximagesink

and moving up into much more complex examples, including one that combined audio and video. GStreamer is based around the idea of sources (i.e. inputs) and sinks (i.e. outputs), along with capabilities (i.e. filters), that can all be combined with a pipe-like syntax using '!' rather than the shell's '|'. One can examine the available sources/sinks/capabilities with:

    $ gst-inspect | more

But the command line doesn't allow programmatic changes, nor changes "on the fly".

For doing more interesting things with GStreamer, Savage pointed to gst-python, which are Python bindings for the GStreamer library. Using gst-python allows adding and removing filters or adjusting audio/video inputs and outputs in real-time under the control of a program. It has been used to build various applications including audio players like Rhythmbox and Amarok.

Savage is quite excited about another, fairly new tool that uses gst-python. Freeseer is an "open source video studio in a backpack", according to its web page, that is built on GStreamer, Python, and KDE. It originally used the FFmpeg command line, but switched over to gst-python after a fairly short time. It is designed for recording conferences, but is more versatile than that.

Freeseer has a simple interface and, as Savage demonstrated, just hitting the "record" button starts recording the screen. It can record from various sources, including audio sources, and is set up to tag the multimedia files with things like which room is being recorded, title and presenter from an RSS feed, and so on. There are some features that Freeseer is missing—streaming, multiple video sources for picture-in-picture, and automatic upload for example—but development is moving quickly so those things, and undoubtedly others, will be addressed soon.

Savage was the executive director for this year's UTOSC, so he set up systems using Freeseer to do the audio and slide recording for each presentation room. He borrowed refurbished, low-end systems from the Electroregeneration society, and added in around $100 worth of equipment (screen grabber and microphone) for each room. He recommended a $50 Radio Shack wireless lapel microphone as the best he had found for presentations, and recording systems only require a Pentium 4 or better. So, for "a few hundred dollars", anyone can put together a system to record presentations, and it wouldn't take that much more to add in live video.

That was clearly a major part of Savage's intent with his presentation: to get other conferences, LUGs, and user groups of all kinds to do more recording of the presentations that they have. From a financial standpoint, it is well within the budget of almost any group, and the free tools are available and getting better. The financial barriers to entry for any group—technical or not—are pretty low, which is quite a change from where things stood five or even three years ago.

---

(Log in to post comments)