Making and processing audio recordings
of conferences and lectures

I am fairly regularly engaged these days to record meetings, lectures and conferences, either so that a written report can be prepared, or increasingly so that the recordings can be put online as MP3 files. Here I recount some of my experiences.

MEETINGS AND CONFERENCES are great opportunities to acquire knowledge and advance our thinking, but inevitably many people can’t attend. If proceedings can be published, many more people will benefit. Even those who were present may be grateful to have their memories jogged!

Academic and professional societies have long made sure this happens. Soon after the Royal Society was founded in 1660 as a London club for the advancement of science, it started to publish accounts of its meetings in the Philosophical Transactions. Today, millions of scholarly lectures are published each year. The system relies on lecturers’ willingness to produce a written paper at their own expense, because of the fame and academic credibility it will give them.

In many other contexts, especially in business, it doesn’t work like that. For example, I’m active around a number of Branches and Specialist Groups of the British Computer Society, and other voluntary associations such as the International Society for Knowledge Organization (UK) and the Network for Information and Knowledge Exchange, NetIKX. These are groups with small budgets, and can’t afford to pay fees to speakers. We’re grateful that these busy people can spare us the time. We can hardly ask them for a written paper as well! If we want to produce a more thorough report of a meeting, we have to do that ourselves.

Because it is hard to take an active part in a meeting and at the same time make accurate notes, I long ago started to make audio recordings of meetings and write them up afterwards. From 2003 onwards, I also started to process audio recordings into MP3 files, which could be posted directly to the Web.

This essay is based on one I wrote for the Swedish periodical Kunskapsbrevet in 2004 but brought up to date to reflect new technology and more recent experience, I’m sharing the techniques I have learned — and the things to watch out for.

Originally this article contained a lot of detail about a range of recording technologies, but I have now transferred those details to a separate article within this Web site, and here I focus more on setting up digital audio recording kit, operating it while the meeting is in progress, and what may be involved in processing the audio files afterwards.

Digital audio and video note-taking

Is your primary reason for recording to help produce a written account afterwards? In that case your main concern is to capture the essence of the event with the least amount of difficulty, and make it as easy as possible to transcribe; the technical recording quality is not so important.

An inexpensive pocket digital recorder with built-in microphones can often do a pretty good job of capturing a recording of a round-table meeting or even a lecture. For a few years I have carried an Olympus DS-50 recorder around with me, and some of the recordings I have made with this have even been good enough to podcast.

There are occasions when a video recording makes a better ‘notebook’ than audio, especially if the presentations have a strong visual component. The Flip digital camcorder is a popular and inexpensive tool, and I have been at meetings where the speaker faced a front row of people pointing their iPhones at him!

When audio quality matters

Capturing really good quality audio becomes more important when the intent is not just to have an aide-memoire for writing something up after, but you actually want to put the lecture up on the Web in ‘podcast’ format. My inspiration for this comes in part from the way the BBC uses podcasting, such as the weekly studio discussion show ‘In Our Time’ hosted by Melvyn Bragg, recordings from which are available as MP3 files from the evening of the day on which the programme was first broadcast, and then stay up there permanently.

Quality recording starts with the microphone, which is to an audio recorder what a lens is to a camera. I long ago decided to graduate to the use of digital audio recorders that can capture the input from one or two good-quality microphones plugged into it — or alternatively, tap the ‘line level output’ from a public address system or a mixer desk.

These days I use a solid-state recorder, a Marantz PMD-661. It records uncompressed WAV files to SDHC memory cards, and one 8-gigabyte card generally gets me through a full one-day conference. Four NiMH rechargeable AA cells usually last at least two hours recording, or I plug in the mains power supply. I explain more about the technical characteristics of recorders and microphones in this companion article.

Getting your microphones in position and in trim

The typical meeting which I record is one in which there is a speaker on stage and an audience who will be given the opportunity to ask questions or make comments afterwards. This can be a challenging brief for the sound recordist. Experience has taught me to be realistic in how I set up my microphones and equipment, to build in flexibility from the start. Ideally I try to capture both the lecture and the Q&A, though sometimes the acoustics makes the latter very difficult.

Primary condenser mic in position

Conrad’s preferred primary microphone is a Røde NT-1A large-diaphragm conderser mic, shown here in a shockmount on a mic stand. It is positioned about 1.2 m from the speaker — not an ‘ideal’ placement, but a pragmatic one, because if the speaker moves away from the mic it will not have as proportionately great an effect as if the mic were placed closer in. The gain level has had to be cranked up somewhat to compensate.

For recording the speaker at the front of a meeting, I prefer a good-quality microphone with a ‘cardioid’ (heart-shaped) pick-up pattern, which rejects sound coming from behind the mic, and suppresses some of what comes from the sides. My Røde NT-1a large-diaphragm cardioid condenser mic has a wonderfully clear sound and the circuits generate very little internal electronic hum, so this would be my preferred recording instrument.

In an ideal set-up, I’d have this microphone placed some 25 cm from the speaker’s mouth, and slightly off-axis. (This ‘off-axis’ trick is to avoid the shock-waves from plosive consonants such as ‘p’ and ‘b’ and ‘t’ from hitting the microphone with a bang.)

The reason why the placement described above is (in theory) ‘ideal’ is that if you can set the gain control fairly low on your recorder, and rely on a good strong noise coming from the speaker, you won’t be amplifying the input too much, so you won’t pick up much background noise from the data projector fan, the air conditioning, the shuffling and sneezing from the audience and so on. Also, even the best mic produces some electronic noise inside its circuits, and you don’t want to amplify that to the point where it becomes intrusive.

However, this works only with experienced microphone users who can be trusted to maintain a proper and constant distance from the microphone as if it were second nature. Less experienced speakers are likely to get too close and bellow into the microphone, or wander off into the middle distance and fade away.

Experience has taught me to place my microphone about a metre away from the speaker, and be more generous with the gain setting. That way it doesn’t cause quite so much variation in the volume if the speaker leans 20 cm closer to the mic or wanders another metre or so further away. I do pick up more background noise with this set-up — but I think it’s a reasonable trade.

(You’ll get away with this compromise better if you use a good-quality microphone, the circuits of which don’t generate as much electronic ‘hum’ internally.)

Riding shotgun

However, this main mic, placed facing the speaker, will do me no good at all when it comes to Q&A. For that, I have a second microphone in reserve, plugged in and switched on, but with the gain set really low on standby. For this role, my choice is a highly directional ‘short shotgun’ mic of the sort often favoured by film-makers. The pick-up pattern of this is more like a cone, and I need to point it in the direction of the person in the body of the hall who is asking the question, or making some point or other.

If I have to travel light, I sometimes pack only the shotgun mic, which I aim at the speaker or at the audience as appropriate. This can raise a dilemma as to where to point the thing, when the to-and-fro between podium and hall gets fast and furious! I also don’t like relying only on the shotgun mic because it has somewhat noisy circuits and doesn’t capture such a clean sound; for the lecturer’s input (which generally matters more) I prefer the clean sound of the condenser mic.

Some speakers just won’t stand in one predictable place so that you can record them easily. Quite often a speaker spends most of the time near the lectern or the laptop, but will from time to time walk over to the screen to point at some part of the projected image. That is where you will see me make a lunge for the gain control for the shotgun mic, which I use as a sound reinforcement for the speaker on the prowl…

Gain without pain!

Recording setup with Sennheiser mic, Audio Buddy and Marantz PMD-661

Conrad’s ‘recording bench’ set-up at an ISKO UK conference in November 2010. A is the secondary mic, the Sennheiser shotgun mic used for tracking speakers as they wander away from the primary mic, and to pick up comments from the audience. It is held in a Rycote shockmounted pistol-grip, with a small tripod screwed into the base. B is the Audio Buddy pre-amp: its two face-mounted knobs make gain level adjustment easy. C is the Marantz digital audio recorder, taking line-level input from the pre-amp.

My first recommendation is — never rely on Automatic Gain Control! This common feature of recording devices is a system that monitors the input from your microphones and sets input amplification (‘gain’) automatically to suit. However, as soon as there is a pause in discussion, the AGC typically loses its nerve and boosts the amplification. The background noises and hisses generated within the recording circuitry get louder and louder, until speaking starts again and AGC retreats to a more reasonable level.

It’s far better to take manual control — monitor the quality of the input with headphones, and keep your fingers hovering near the manual gain control knobs on your recording apparatus, ready to jump in and make appropriate adjustments.

When I am ‘on the job’ at a conference, I’ll be wearing a pair of closed-back monitoring headphones and listening carefully throughout. However, I have learned not to rely on these as an indicator of whether the gain settings are trimmed appropriately: after all, the gain could be set too low and the headphone output too high, or vice-versa. I keep my eye on the visual gain-level display, trimming it so that I’m amplifying the inputs generously but not peaking too high, which would cause a distortion.

If you use two mics as I prefer, you should be able to control the gain for each independently of the other, easily and smoothly. This is one of the shortcomings of the Marantz PMD-661: it has independent left and right recording level controls, but they are concentric, and hard to adjust relative to one another, certainly with one hand while the other is aiming a shotgun mic at the audience!

An external microphone pre-amplifier can help here. This sits between the recorder and the microphones, provides the mics with phantom power if required, and gives me smooth independent adjustment of the gain levels for the the microphones. What comes out the back is line-level input, so I switch the Marantz XLR inputs from ‘mic’ to ‘line’ and I am back in full control.

For a few years my external pre-amp has been the ‘Audio Buddy’ shown in the photo above right. In time for the ISKO UK conference in July 2011, I purchased a small Alesis mixer desk which lets me mix the output from up to four microphones and gives me extra controls such as three-point equalisation.

Getting set up (and torn down again!)

When I’m asked to record at an event, it is often the case that I don’t have prior sight of the venue. It can take a while to get set up, so I turn up early. I usually operate from the front row of the lecture hall, taking a seat at one end of a row next to an aisle, firstly because I am often doing double duty as event photographer, secondly because I then have more freedom of movement to rise to my feet at Q&A time with the directional shotgun mic in my hand.

If I’m using the Audio Buddy or the Alesis mixer desk, I need access to a power point. I have learned to lug a 30-metre power extension cable reel with me, as in some older lecture theatres one can find oneself far from a source of electricity! In some lecture halls I’ll find a shelf or table to set out the recording deck and the pre-amp, but as a back-up I carry a folding camping stool as a work surface.

I carry a number of XLR cables with me, and they can be plugged end to end to lay a cable from the main mic to my recording gear. I confess to being quite nervous about cables because they are prone to internal rupture, so I usually carry more than I need, and check I’m getting a signal through well in advance of the start of the meeting. I also carry a roll of ‘gaffer tape’ so I can tape down XLR and power cables — I don’t want to be the cause of someone tripping over…

One of the penalties of being the sound recordist can be that at the end of the meeting, people disappear off to the pub and you are left behind still clearing up! Or if there is a mid-event refreshment break, I may be left behind in the room, guarding my set-up.

Unwanted shocks and disturbances

Then there are the unwanted intrustions which can interfere with the quality of the recording. At University College London, which is the usual venue for ISKO-UK meetings, there have been some ongoing rebuilding works, with jackhammers and angle grinders competing with the speaker. Some of the larger, older lecture halls also have hard acoustics with a pronounced echo, ensuring that every cough and sneeze from the audience will make its way into the recording.

One thing that one can and should guard against is noises getting into the microphones via the mic stand, caused by people walking across the floor or (if the mic is on a table stand) knocking the table. I always mount the studio condenser mic in a ‘shockmount’, which has a lattice of rubber bands to suspend the mic body. I also tie up the cable to the mic stand, leaving a small, loose loop, as a knock to the cable is another route by which vibrations can enter the body of the mic.

On one occasion, in a large conference hall at The Open University in Milton Keynes, we had the benefit of a public address system, and mounted on the lectern was a discreet small version of a shotgun mic, the AKG C747 MkII. These are beautiful little speaker mics, and not cheap at nearly £500, but one flaw they have is that the thin cable that runs from them is not shielded against radio frequency interference! Time and again, the public address system was assailed by a kind of ‘chuntering’ bleat, as the speaker’s mobile phone sought to re-establish contact with the phone network. We asked again and again for people to switch off their mobile phones — but all they did was switch them to silent mode, which did nothing to solve the radio interference problem!

Editing audio for distribution

If the audio recording is intended for wider distribution, usually as a downloadable MP3 file, my next job is to edit it. The audio files are transferred to the computer, and I use Apple Quicktime Pro to do a first rough trimming of the files to length, and convert to an uncompressed audio format (I prefer AIFF). Then I open the files in Apple Soundtrack Pro and start on the edit proper.

Without a moving picture to give the game away, you can remove all sorts of snippets and chunks from an audio recording, and no-one will be the wiser! (Radio editors do this all the time.) Sometimes it is politic to remove from the public record a contentious statement… but usually the intent of the editing is simply to tighten up the recording, reducing the play time and file size.

It is amazing how much can be removed from the average talk without loss, and with some improvement. ‘Ums’ and ‘ahs’ and false starts can be clipped out, hissing intakes of breath suppressed, and each pause can have a fraction of a second sliced from it. Often a thirty-minute talk can lose five minutes of this stuff, without removing any real content.

If the lecture hall was small and had good acoustics, and the shotgun mic was able (with a few twiddles of the gain control knob) to bring distant contributors into range, the left and right channels can often be simply merged into a combined mono channel, which reduces file size considerably. On occasion, where I've had to capture questions from thirty rows back in the auditorium, from someone with a small voice, I’ve needed to adjust the amplitude of the quieter channel in post-production before doing a merge to mono, which adds considerably to the edit time.

On the whole, though, the edits and tweaking take about four to five times the length of the recording for me to accomplish. In other words, for a talk of 30 minutes followed by 15 minutes Q&A, I can expect to have to put in three to three-and-a-half hours of editing and ‘sweetening’.

Just occasionally, I need to put something into the edit, rather than just take stuff out. At two recent talks at NetIKX meetings, the speaker referred to something like ‘this stuff on the screen, which you can read for yourselves, so I don’t need to read it out’. Fine for the audience present… not for those who only get to hear the audio afterwards. On those occasions, I recorded a ‘patch’ along the lines of — At this point, the speaker showed a list of factors which are important to remember. She did not read these out, but they were… and then narrate what was on the screen, and insert that at the appropriate point in the recording.

Conversion, compression, distribution

The first time I got involved in preparing audio files for posting as MP3 files on the Web was following a one-day conference of the British Computer Society’s Electronic Publishing Specialist Group, and event about copyright and the Internet. We found ourselves with a fine collection of recorded talks which worked well as a purely audio experience, and resolved to post them online.

A typical talk from that event, as a 16-bit stereo file with a sampling frequency of 44.1 kHz, was about 480 megabytes of data: not something you’d want to download. First I reduced each recording to mono and halved the sampling frequency to 22.05 kHz. This reduced the filesize to a quarter, with only a small loss of quality.

Finally I experimented with increasingly severe levels of MP3 compression. MP3 is a ‘lossy’ compression method. At the stronger levels of applied compression, the recording definitely becomes more ‘swooshy’ and hollow-sounding, but one could still make out every word. Setting an upper average data limit of about 6 kilobits per second, and using iTunes for MP3 conversion, I produced a series of files with an average size of 3 Mb per talk.

In 2003 there were still many people who did not have access to broadband. These days, I do not apply such severe compression to files before uploading them. I typically choose a variable bit rate (VBR) MP3 compression method with an average data throughput of 56 kbps, mono — about the same as the settings used by the BBC for its podcast documentaries and talk radio shows.