Digital Recording of House of Worship Services

This is the first of a series of articles about contemporary digital recording practice in a volunteer church service recording context. The advent of digital mixers and digital audio workstations has completely changed this game allowing a patient amateur to make a clean pleasing service recording for distribution to shut-in and traveling members.

Those wishing to learn more about copyright in house of worship program production are advised to see reference 1 for a complete discussion. Copyright law specifically exempts houses of worship from royalty payments for material performed or presented as part of a worship service proper. Houses of worship are required to pay royalties for non-worship use of copyrighted material not in the public domain. For this reason, we distribute service recordings to members on physical media and include only material copyright by the church or our speaker in our weekly sermon podcast that is publicly available.

After the break, we’ll talk about track capture off of a digital mixer. Although not hard, each mixer has its quirks and limitations so our intent here is to alert you to typical challenges and limitations rather than to provide switchology for a specific mixer or mixer family.

Revision History

2019=08-05 Update for lessons learned since the original draft

References

The following references are provided for those desiring to learn more about products commonly used for track capture and track mix-down. Our church happens to have an Allen-Heath SQ-5 mixer and we use Studio One demo version because it is free, is available for MacOS and Win10, and the feature set is adequate for splitting a service recording into service parts tracks and mixing the tracks down to 2 channel stereo for distribution to congregants.

Over the year that we have used our Allen & Heath SQ5 mixer, we have installed a number of software updates. With each update, we have seen the behavior the mixer as a track capture device improve as A&H corrected issues in the USB subsystem and disk I/O subsystem. The material presented here reflects Release 1.4.1 of the mixer’s firmware.

Motivation for Multitrack Recording

We originally made service recordings by sending a mixer Aux output to a Griffin iMic on a Mac Mini running Rogue Amoeba Audio HiJack. Well, 15 years on, Griffin no longer makes iMic or supports it but it is core audio compliant and just works. This worked OK for spoken word but not music. Then, we moved, and installed state of the art digital audio in new hall. Finally, we could play a recording without embarrassment (well some program material but not all).

The Mac Mini is pre-2009 and still has MacOS Lion installed so it cannot run a current DAW. Its libraries are too far behind MacOS Mojave and Apple has added libraries since Lion was feature complete. Because of its age, this Mac cannot be updated to Mojave so we have to retire it or run Linux on it. (MS Windows and audio production don’t mix). With our new mixer, the HiJack captured audio is too low in level to make a good recording.

FOH Engineer Practice Varies

Taking a stereo digital capture of the mains, we found was that the spoken word bits of the service were clear but performed music was not always clear. Instruments would be missing or out of balance.

We were not always miking the piano, drums, or electric guitars so, though the hall sound was decent in our small worship space, the recorded sound was a mix of close miked vocals and back-scattered instrument sound coming on stage after reflection off the hall back wall and stage front wall. So instruments were delayed and down.

When we were miking the instruments, the engineer was mixing for hall sound as heard from a mix position at the back of the audience under the balcony. Most engineers were not using headphones to evaluate the balance of the mix fed to the PA. Often, the house sound would be good because the PA was supporting the instruments but the recording would have muted instruments as direct sound allowed a reduced level in the PA mix.

The New Mixer

It’s a digital mixer with 16 channels of input and 8 channels of output at the mixer desk and 16 channels of remote inputs and 8 channels of remote outputs on stage. We use the local channels for the wireless microphone inputs and the local outputs for the assistive listening system feed and the lobby sound feed. We also send an aux output to the old Mac.

The feed to the mains is digital and powered speakers apply time delay correction and device equalization and hall equalization internally saving us a drive rack and external DSP equalization and time alignment processors. We also minimize conversions between analog and digital further reducing system cost and maintaining clarity.

The mixer, like many, can send selected channels to USB in one of two formats, audio interface device streaming format or disk block device format via the file system. Our mixer writes up to 32 tracks of output to individual WAV track files. We can patch selected processing channels to the USB output stream.

The patching can be at the preamp output or the channel strip output or from one of the mix bus outputs. you set this up on the I/O page of the mixer using its touch screen. Any source can be routed to multiple tracks by making multiple entries in the source patching row.

The mixer has a bit of personality

As we’ve worked with the mixer, we’ve realized that it has a bit of personality. By personality, I mean that the mixer is smarter than the tech configuring it. It is smart enough to figure out if a given channel has a source connected. I don’t know if it does this using some plug insertion sensing or signal presence sensing. But the mixer ignores open inputs that are in the record channel list.

More interestingly, it appears to process inputs until it finds an open input and ignores the remainder of the list because nobody in their right mind would randomly patch record outputs. Well, we do ….

What we found was happening is that our performers change week to week. Some weeks we have choir and some weeks not. Some weeks we have soloists and some weeks not. Some weeks we have guitar and drums and some weeks not depending on what our volunteers want to play and how our half-time music director allocates his time between his day jobs, VSO chorus, and teaching. So the mikes on stage change each week.

So, we’d record a service, take the tracks back home, and find some with program material, some zero, and some noise. Then in an update, our mixer vendor changed USB stream behavior to stop at the first open channel in the patch list. Active mikes in the patch list following the open channel were not being written to the stream.

Recommendations for track patching

So with this experience in mind, we can make the following track patching recommendations. These are based on the behavior of our mixer and the varying ability of our amateur front of house mix engineers who may also tend the stage lights and video projector. The intent is to be able to take a Mulligan on any service element not properly mixed to the mains.

Patch the main outputs to the first two tracks
Patch pulpit mike, lectern mike, and handheld mikes that are always used.
Determine the active stage mike channels and patch those actually used each week. Don’t leave unused channels in the track output list. (This turned out not to be a problem)
Always patch the preamp output after digitization but before channel processing. Many mixers have taps at preamp, before fader, and after fader. It is best to grab signal before mute, filtering, compression and limiting, and mix level controls.
Patch 16 or less outputs. Patching #17 will double the disk space used as the mixer will start a second set of 16 track files (we assume this as we have not gone beyond 16 patches).

Lessons Learned about Track Patching

We found that our mixer was going to write 16 tracks to disk whether all 16 were patched or not. When a track was not patched, we found the track full of dither noise. When a track was patched, there was always audio because we were tapping the processing channels between the digitizer output and the start of channel processing (the tie line taps). This is what we intended so we could re-equalize and remix after the service.

We found that the following scheme makes it easy t remember how the recording channels are mapped by keeping the stage channels directly mapped to the corresponding USB channel. This makes it quicker to locate a missing patch.

We found that we could take recording and video playback audio from the main output so we use the main for anything we forgot to patch.

We found that we occasionally place a condenser microphone without energizing its phantom power. This is kind of hard to recover.

Use the mixer’s iPad app to patch inputs to channels and outputs to jacks and to the recording outputs (usually USB track channels or Dante track channels). The iPad app has better known touch characteristics.
Confirm that the correct input group tab and correct output group tab are selected before touching the patch matrix. This will reduce the likelihood of a stray patch change occurring away from where you intended to make patch changes.
Patch the one of the main outputs (mono) or both (stereo) and the wireless channels at the bottom of the patch matrix. Patch the stage channels at the top of the patch list.
Begin patching the stage channels at the top of the patch matrix. Our stage boxes are patched so that channels 1 to 6 are commonly used. This scheme keeps these channels directly mapped to recording channels one to six.
Our piano, piano vocal, and choir mikes are on channels 13 to 16. The down stage left and right boxes are channels 7-9 and 10-12 on the remote I/O.
Our wireless system is on the mix desk local I/O

Lessons Learned about Disks

Along the way we learned a bit about selecting disk media for stream capture and preparing disk media for recording.

SSD vs HD

If it has moving parts, it will break. Any HD will probably die as a result of being carried about between church and home. Mobile computing equipment breaks! Always! One too many bounces, shakes, or bungie jumps and kaput!

SSD write speeds vary wildly from brand to brand and within a brand. Because of standards and branding, we have found it best to use either 64 GB uSD cards in an adapter or a purpose built SSD like those from Glyph.

A 250 GB Glyph Atom disk can be had for $100 or so and will hold a month’s worth of track files. The Glyph Atom disks are designed specifically for use by a DAW or mixer as audio recording media. USB interface, storage control chip sets, and flash chips are specified for the recording application. The Glyph Atom will write at USB 3.1 Thunderbolt line speed.

Generally, we have found that uSD cards intended for use with interchangeable lens still/video cameras are fast enough to use for mixer tracking devices. Be very careful to buy these from authorized distributors as counterfeiting is rampant in China and eBay or Amazon store front sellers may be selling under performing or fraudulently labeled product.

USD cards are about $15 and can hold a single service. Because of their small size, they are hard to carry about and can go missing from a pocket. Altoids tins are very useful. Five are equivalent to the Glyph device in capacity.

Sizing the Disk

How much disk space is needed? In multi-track mode, our mixer always writes 16 track files if 16 or less tracks are patched to sources. We’re finding that a 1 hour track is about 1.5 GB of 24 bit audio at 96 k sampling rate. So a one hour service should fit on a 32 GB uSD card. Sermons never run over but complete services commonly go into extra innings as each of 15 or so service elements runs a tad long.

So we recommend a minimum of 64 GB of free space on the recording media. Thirty-two should be plenty but 64 GB allows the mixer to write a full 32 tracks for a 2 hour service.

Note that FAT32 maximum file size is 4GB so 2 hours is a practical time limit for a take.

Preparing a Disk

But our mixer formats nothing bigger than 32 GB. We found we could format larger disks using MacOS Disk Utility picking MBR for the boot block type and FAT32 or ExFAT (long names) for the disk type. The mixer will accept a properly pre-formatted disk larger than 32 GB but we don’t know if it will use the whole disk.

The mixer’s format operation also puts the expected file directory hierarchy on the media so you must do the same. Fortunately, there are no hidden files required so it is sufficient to copy over the empty directories or hand-create them with care.

Takes longer than two hours

The FAT32 file system used by the mixer has a 4 GB file size limit. This means that recording a long service or a long event will require multiple sets of track files. For our mixer, this would mean stopping the current session and starting a new one. This can happen at a break between program elements.