1. DocLing 2016 David Nathan & Anthony Jukes Audio theory and practice for language documentation

2. An epistemology for audio in documentation  an audio recording is made in order to be experienced by a human listener  a recording conveys what a human listener would experience at a particular location in an event setting  documentation goals define recording methodology  a recording should capture spatial information  metadata about the recording and the recording setting are required for full interpretation  ethical recording respects speakers and honours their contribution through your effort and skill

3. Evaluating recordings  accuracy: how well is the signal captured, as true to its sources and without distortion?  intelligibility/information accessibility: can the desired content be identified?  signal vs. noise: is the ratio acceptable? can the focal source be separated from all sources of noise?  listenability/comfort/aesthetics: is it easy on the ears? will it be comfortable to listen to for an extended time?

4. Evaluating recordings  localisation of sources: is enough spatial information captured?  separation of noise: can all sources of noise be separated?  representation of environment: are the acoustic properties of the recording space appropriately represented?

5. Evaluating recordings  content (identity, performance, uniqueness, coverage): were the right people recorded doing the right things?  editability/repurposeability: is the recording suitable for turning to relevant purposes?

6. Recording audio  making it is both art and science  a critical and ethical responsibility  strongest relationship to communities  it’s not necessary to record everything, but it is necessary to record well

7. SIGNAL & NOISE

8. Evaluating recordings  signal  noise  signal to noise ratio  listenability (eg comfort, consistency)  fit for purpose

9. Evaluating recordings  audio professionals use their human ears as evaluator of audio quality and value, while many linguists (mistakenly?) look to formats, spectrographs, wave- forms, analyses etc  44.1 KHz, 24 bit

10. Signal - what you want  content  contextual and spatial information  fidelity  comfortable to listen to

11. Noise - what you don’t want  from environment:  near: people, animals, activities  far: traffic, generators, planes  machines: refrigerators, fans, computers  not hearable: mobile phones, electrical interference  acoustic: reflections/resonance

12. Noise - what you don’t want  generated by unwanted parts of event  shuffling papers, clothes  table banging  backchannel from interviewer  equipment handling, especially microphones and cables (and recorders with built-in mics)

13. Avoiding handling noise  use stands and cradles etc

14. Noise - what you don’t want  generated by equipment  wrong input levels  circuity noise (cheap or incompatible)  compression loss or distortion  ALC/AGC effects (pumping)  video camera motors

15. External noise sources examplepossibilities for dealing with it trafficinvestigate, record in quiet time face away use damping materials childrenget them involved show something to satisfy curiosity animalschoose time of day weather (wind, thunder, rain etc) use dead cat; wait; reschedule

16. Dead cat

17. Close-up noise sources  machines examplepossibilities for dealing with it refrigeratorpre-survey what comes on intermittently turn off relocate motors, switchingmonitor fansmonitor, dead cat (windshield)

18. Dealing with noise sources  be prepared and aware  seek collaboration  monitor  use or modify room acoustics  location  direction  surfaces  reflection  absorption  isolation

19. Utilising room acoustics  location  away from doors, windows, traffic areas  direction  face away from noise sources  reflection  avoid parallel surfaces  surfaces  avoid hard smooth surfaces  choose or create soft or rough surfaces  isolation  find an ‘’airtight’’ place

20. When is a noise not a noise?  When it is part of the content, for some interpretation of the event Performance of John Cage 4”33’John Cage Available on iTunesAvailable on iTunes (150 yen)

21. PERCEPTION & PSYCHOACOUSTICS

22. Audio perception/psychoacoustics  a human listener has:  location, orientation in a physical setting  two ears - incredibly sensitive  a brain/mind  the mind selects from various sources of sound and other sensory information, using long- and short-term memory  listening is actually a “hallucination”

23. Psychoacoustics and recording  microphones don’t have a mind: they can't distinguish wanted from unwanted sound  microphones don’t have “edges” like camera lenses

24. Psychoacoustics and recording  the recording process loses acoustic information  if you only care about transcription, then you are going to throw away over 99% of the acoustic information anyway! real world record acoustic phenomena represent (some) linguistic components derive data

25. Implications for recording  typical recording methods are unscientific!  … so what should we do?

26. Implications for recording  plan and manage recording  goals  equipment preparation and settings  other preparation  environment and setup  sources  changes, actions, settings

27. Implications for recording  why is it important to record spatial information?  what other information (acoustic or non-acoustic) do we need?

28. “Sound stage”  spatial information is an essential part of audio  we are amazingly attuned to it  we should record in stereo

29. “Sound stage” ... or ORTF (binaural)

30. MICROPHONES

31. Microphones and audio quality  microphones are the greatest factor in audio recording quality  selection of appropriate microphone(s) for the task  placement and handling

32. Microphone types  principle: dynamic vs condenser  directionality: omni, cardoid, and shotgun  spatiality: mono, stereo, ORTF, binaural

33. Microphone physical principles  dynamic  generate signal from sound pressure  more robust, less accurate  used for musical and live performance  condenser  more fragile, sensitive and accurate  need power source - battery or phantom power  in general, use condenser microphones for language documentation

34. Omni  lavalier or tie-clip microphones are typically omni- directional

35. Microphone directionality - omni omni-directional

36. Cardioid  many “standard” handheld microphones are cardioid units

37. Microphone directionality - cardioid cardioid

38. Shotgun (= directional, hypercardioid)  shotguns are good for  quiet sources  in some noisy environments  video work

39. Microphone directionality - shotgun shotgun/directional/hypercardioid

40. Head-mounted microphones  head-mounted microphones are excellent for very noisy environments or mobile activties, and may be omni- directional or cardioid

41. Stereo microphones  spatial information is an important part of audio

42. Full “sound stage”: ORTF Superlux S502 Full binaural on dummy head

43. ORTF & Binaural ORTF is now the “best practice” for field recordings” (Austrian Academy of Science, Vienna Phonogrammarchiv)

44. Simulating ORTF with 2 cardioids 17cm 110°

45. Microphones - quality  generally, you get what you pay for  each model has its own subjective “colour”  good microphones for language documentation cost from US$180 to US$500

46. Reputable makers - examples  AKG  Audio Technica  Beyerdynamic  Røde  Sennheiser  Shure  Sony

47. Microphone placement

48. Microphone usage principles  where should the microphone be?  in general, about 20cm from the speaker’s mouth  the inverse square law is your friend...

49. The inverse square law

50. Using the inverse square law  if you have noise sources, increase the signal to noise ratio by:  placing the microphone as close as possible to the signal source  placing the microphone as far as possible from the noise source

51. Microphone connections  plugs  cable types  cables for stereo/mono, multiple  wireless  power sources for condenser microphones - battery or phantom power  see http://www.hrelp.org/archive/advice/microphones.html

52. Microphone connections  1/8 inch “mini-jack”  RCA/ “phono plug”  1/4 inch “jack plug”  XLR “Canon”

53. XLR  the physical connection is independent of the electrical connection  transmits phantom power  low-noise over long cable runs  you can use XML to mini-jack cables or converters for recorders with mini-jack inputs

54. Choice of Recorders  Connectors - XLR or mini-jack or both?  Ruggedness and build  Accuracy  Media type  Battery life  Cost

55. AUDIO WORKFLOW

56. Audio workflow who/what/where /why/how do you want to record? contact people audio training budget, research, and buy equipment assemble, test, practise Before you go

57. Audio workflow transport safely check environment, situations, permissions make test recordings local training & collaboration On site, before recording

58. Audio workflow record! monitor! collect metadata check quality monitor Sessions select equipment (microphones)

59. Audio workflow (label)check quality backupadd information (metadata, metadocumentation, transcriptions, annotations, etc) After sessions

60. Audio workflow add information (metadata, metadocumentation, transcriptions, annotations, etc) package and send to archive Later other audio outcomes and resources

61. End !