1 Language Documentation in West Africa July 19 2010 Winneba, Ghana David Nathan Endangered Languages Archive Hans Rausing Endangered Languages Project SOAS, University of London Audio Recording Techniques & Equipment
1
Language Documentation in West AfricaJuly 19 2010
Winneba, Ghana
David NathanEndangered Languages Archive
Hans Rausing Endangered Languages ProjectSOAS, University of London
Audio Recording Techniques & Equipment
2
Topics - session 1
Questions Audio workflow Evaluating recordings Perception and psychacoustics Microphones Connections Recorders Carriers
3
QUESTIONS
4
Big questions
What are we actually recording? What is it for? What is the role of audio in language
documentation?
5
What is audio?
Audio is not data real world record phenomena represent phenomena derive data
Audio is a resource 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
neceessary to record well
6
AUDIO WORKFLOW
7
Audio workflow
who/what/where /why/how do you want to record?
contact people
audio training
equipment & budget
assemble, test, practise
Before you go
8
Audio workflow
transport safely
check environment, situations, permissions
make test recordings
local training & collaboration
On site, before recording
9
Audio workflow
record!
monitor!
collect metadata
label check quality
monitor
Sessions
10
Audio workflow
label check quality
backup add information (transcriptions, annotations, metadata etc)
After sessions
11
Audio workflow
send samples to archive
add information (transcriptions, annotations, metadata etc)
... package and send to archive
Later
12
EVALUATING RECORDINGS
13
Evaluating recordings
signal noise signal to noise ratio listenability (eg comfort, consistency) fit for purpose
14
Evaluating recordings
audio professionals use the human ear as evaluator of audio quality and value, while many linguists mistakenly look to formats, wave-forms, analyses etc
44.1 KHz, 24 bit
15
Signal - what you want
content contextual and spatial information fidelity comfortable to listen to
16
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
17
Noise - what you don’t want
generated by event (unwanted) shuffling papers, clothes table banging backchannel from interviewer equipment handling, especially
microphones and cables
18
Avoiding handling noise
use stands and cradles etc
19
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
20
Evaluating environment/situation
external environment access electricity external noise sources
21
External noise sources
example possibilities for dealing with it
traffic investigate, record in quiet time
face away
use damping materials
children get them involved
show something to satisfy curiosity
animals choose time of day
weather (wind, thunder, rain etc)
use dead cat; wait; reschedule
see also General principles
22
Dead cat
23
Close-up noise sources
machinesexample possibilities for dealing with it
refrigerator pre-survey what comes on intermittently
turn off
relocate
motors, switching monitor
fans monitor, dead cat (windjammer)
24
Dealing with noise sources
be prepared and aware seek collaboration monitor use or modify room acoustics
location direction surfaces reflection absorption isolation
25
Room acoustics
location away from doors, windows, traffic areas
direction face away from noise sources
surfaces avoid hard smooth surfaces
reflection avoid parallel surfaces
absorption choose or create soft or rough surfaces
isolation find an ‘’airtight’’ place
26
PERCEPTION & PSYCHOACOUSTICS
27
Audio perception/psychoacoustics
audio information is diverse a human listener has:
location and orientation in physical world two ears - which are incredibly sensitive a brain/mind
the mind merges and selects from various sources of audio information
listening is actually a “hallucination” so what should we record? typical recording methods are unscientific!
28
Psychoacoustics and recording
microphones are not like camera lenses they don’t have “edges” don't distinguish wanted and unwanted
info the recording process removes some
information
29
Implications for recording
you need to set goals, plan and manage recording goals equipment sources environment settings
example: recording spatial information why is this important?
30
“Sound stage”
spatial information is an essential part of audio
we are amazingly attuned to it we should record in stereo
31
“Sound stage”
... or in ORTF (binaural)
32
MICROPHONES
33
Microphones and audio quality
microphones are the greatest determinant of audio recording quality selection of appropriate microphone(s) for
the task placement and handling of the
microphone(s)
34
Microphones in the digital era
microphones in the digital era recorder quality has increased but prices
decreased microphones have become comparatively
more expensive why? microphones are analogue devices!
35
Microphone types
principle: dynamic vs condenser directionality: omni, cardoid, and shotgun spatiality: mono, stereo, binaural
36
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
37
Microphone directionality - omni
omni
38
Omni
lavalier or tie-clip microphones are typically omni-directional
39
Microphone directionality - cardioid
cardioid
40
Cardioid
many “standard” handheld microphones are cardioid units
41
Microphone directionality - shotgun
directional/shotgun/hypercardioid
42
Shotgun
shotguns are good for quiet sources, in some noisy environments, and for video work
43
Stereo microphones
spatial information is an essential part of audio
44
Full “sound stage”: ORTF
45
Simulating ORTF with 2 cardioids
17cm
110°
46
Microphones - quality
generally, you get what you pay for each model has its own subjective colour decent microphones for language
documentation fieldwork cost from £120 to £300
47
Reputable makers - include
AKG Audio Technica Beyerdynamic Røde Sennheiser Shure Sony
48
Microphone placement
49
Microphone usage principles
where should the microphone be? in general, about 20cm from the speaker’s
mouth
the inverse square law is your friend ...
50
The inverse square law
51
The inverse square law
52
Using the inverse square law
if you have noise sources, maximise 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
53
CONNECTIONS
54
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
55
Microphone connections
minijack/miniplug (fragile)
RCA/phono
1/4 inch (headphone)
XLR (Canon)
56
XLR
professionals always use them electrical contact is independent of the
physical connection latching is independent of the electrical
contact
you can use XML-to-miniplug cables or converters for recorders with miniplug inputs
57
RECORDERS
58
Recorders
types and their strengths/weaknesses/implications
quality parameters accuracy (frequency response,
distortion, s/n ratio) reliability features versatility power sources, battery type
and battery life
59
Recorders
media types, costs, properties, implications connections formats
60
Using recorders
settings – levels, formats, AGC/ALC a second recorder? do you have to do it yourself?
61
CARRIERS
62
So you’ve recorded something?
carrier types to label ... or not preservation track the content
you may need to digitise/redigitise/capture it
63
General guidelines for success
microphone choice monitoring familiarity and skill with equipment power and batteries a range of equipment, not the “perfect item”! consistency principle
juxtapositions efficient field sessions and later processing
64
END OF AUDIO BASICS!
65
PART 2: AUDIO PROPERTIES
66
AUDIO SIGNALS
67
Audio is initially analogue
analogue means an infinitely variable property of the real physical world
digital means a sequence of measurements of real world properties, ie symbols
68
Audio signal parameters
pitch kHz - human voice fundamental 100 (m) – 200 (f) Hz formants 800 Hz – 4+ kHz harmonics, other, up to 15 kHz
amplitude (power) dB a relative and logarithmic measure 0 dB is reference point; sound of mosquito
flying at 3m max human is about 140 dB (pain = 120) each 6 dB step perceived as
doubling/halving volume
69
signal to noise ratio of wanted to unwanted sound data the bigger the number the better
Signal parameters
70
DIGITAL AUDIO
71
Digital audio
Analogue Digital (identify and measure points)
72
0 20 40 60 80 100-100
000
1000
0
nominal time
ampl
itude
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
Digital audio
73
Digital audio parameters
digital means measuring or “sampling” where and when is it done?
the properties of digital audio are: sampling rate(Hz) sample size (“resolution”, “bit depth”) mono or stereo for compressed data: bit rate (Kb/s)
74
Digital audio parameters
what do these mean? 11KHz, 8 bit 44.1 KHz, 16 bit 48 KHz, 24 bit 192 KHz, 48 bit
these have implications for quality file size compatibility, usage ...
75
Encoding
“codecs” file formats eg WAV, AIFF, AU, MP3, Ogg
76
reasons types
open and proprietary formats (eg MP3 vs ATRAC)
lossy and non-lossy (most are lossy) repeated compression unpredictable
distinguish sound information content from its encoding and its carrier
Compression
77
Digitising
where is it actually done? involves either
digitisation (capturing/ingesting) re-digitisation (capturing) copying (may involve transcoding, e.g.
ATRAC)
78
Digitising
where was your audio digitised?
79
Digitisation: results and quality
what does the result depend on? player and digitising devices settings levels cables, connections, environment
80
Digitisation: results and quality
where can quality be lost? (as well as original recording issues) poor treatment of carriers unknown properties of carriers (eg unlabeled) choice of output port, settings (level, format
etc) choice of input port, settings etc quality of player and digitising devices connections/cables, interference from other
devices or mains supply
81
End