Perspective

Everyday Listening Situations

Method

Average Speech Levels

and Spectra in Various
Speaking/Listening
Conditions: A Summary
of the Pearson, Bennett,
& Fidell (1977) Report
Wayne O. Olsen
Mayo Clinic, Rochester, MN

n 1977, Pearsons, Bennett, and Fidell

completed a report for the U.S. Environmental Protection Agency describing their
measurements of speech levels in a variety of
settings. Their report, entitled Speech Levels
in Various Noise Environments, Report No.
EPA-600/1-77-025, was prepared for the
Office of Health and Ecological Effects,
Office of Research and Development, U.S.
Environmental Protection Agency in Washington DC.
Pearsons, Bennett, and Fidell collected a
unique yet large sample of data on background levels and the levels of conversational
speech in schools, homes, hospitals, department stores, trains, and airplanes. Their data,
therefore, provide vital information regarding
speech levels and signal-to-noise (S/N) ratios
encountered in everyday listening situations.
In addition, they measured speech levels and
speech spectra of females, males, and children
uttering a standard phrase at various vocal
efforts in an anechoic chamber, thereby
documenting differences in speech spectra
related to gender, age, and vocal effort. The
intent of this paper is to summarize their report
and provide its important information to a
larger segment of the professional community.
Their measurements in everyday listening
environments and in the anechoic chamber are
treated separately in this summary.

Pearsons et al. (1977) completed measurements of speech levels of teachers in 20 classrooms, and for a listener in conversations with
residents inside and outside 25 homes in urban
and suburban areas, with patients and nurses in
23 locations in four hospitals, with personnel in
seven large department stores, with 11 passengers on the Bay Area Transport trains in San
Francisco, and with 12 passengers in four
commercial jet aircraft and one commercial
propeller-type airplane.
In the classroom setting, the teachers speech
was recorded at a lavaliere microphone worn by
each teacher and with microphones at a distance
of 2 m (near the front of the classroom) and a
distance of 7 m (at the back of the classroom).
Measurements from the lavaliere microphone
were mathematically adjusted (i.e., normalized)
to levels equivalent to those that would have
been observed at a distance of 1 m.
For measurements in the home settings,
hospitals, department stores, trains, and
airplanes, recordings were made for the
listener, who wore a microphone near the ear
in an eyeglass frame. Several segments of
continuous conversation at least 10 s in length
by the talker, that is, without responses by
the listener, were recorded. The distance
between the talker and the listener generally
was 1 ma distance voluntarily selected in
the home environment as a usual communication distance. In the train and airplane
settings, the distance between the talker and
listener was approximately 0.5 m. Also,
recordings were completed for the background
noise levels when there was no conversation
between the participants.
Recordings were analyzed with a real-time
one-third octave analysis system. Levels of the
speech and background noise were given in Aweighted sound pressure levels. The integration
time on the analyzer was equivalent to fast on
sound-level meters.

Results
The means and standard deviations for the
background noise and teachers speech levels
while lecturing are provided in Table 1. Mean
background levels in the two schools were 48
and 51 dBA; mean speech levels near the front
of the classroom (2 m) were 62 and 66 dBA, for
schools 1 and 2, respectively, maintaining a
S/N ratio on the order of +15 dB. The speech
level near the back of the classroom (7 m) was
approximately 5 dB weaker. The authors also
reported that, normalized at a distance of 1 m,

American Journal of Audiology Vol. 7 1059-0889 American Speech-Language-Hearing Association

Olsen

TABLE 1. Mean speech and background levels in

dBA for 20 classrooms in two schools. Standard
deviations in ( ).

School #1
School #2

TABLE 3. Mean speech and background levels in

dBA in hospitals and department stores. Standard
deviations in ( ).

Speech

Background
Noise

1m

2m

48 (2)
51 (2)

69 (4)
73 (4)

62 (5)
66 (5)

Speech
7m
57 (4)
62 (6)

Note. All values are rounded to the nearest dB. From

Table II in Pearsons et al. (1977).

Background
Conversation
Noise
1m
Distance
Hosp. Patients Room
Hosp. Nurses Station
Department Store

45 (2)
52 (5)
54 (3)

56 (2)
56 (3)
58 (3)

55 (1)
57 (4)
61 (3)

Note. All values are rounded to the nearest dB. From

Table II in Pearsons et al. (1977).

the teachers speech levels increased approximately 1 dB for each dB increase in noise level
from 45 dBA to 55 dBA.
The average noise levels outside and inside
urban homes were 61 and 48 dBA, respectively;
for suburban homes, the levels were 48 and 41
dBA, outside and inside, respectively (see
Table 2). Average speech levels (normalized to
1 m) were 65 dBA outdoors and 57 dBA
indoors for urban homes. The average speech
levels were 55 dBA both inside and outside for
suburban homes. The S/N ratios at conversation
distances were on the order of +5 dB and +9 dB
outside and inside, respectively, for urban
homes, and +8 dB and +14 dB for like settings
in suburban areas. Pearsons et al. (1977) also
noted that the level of speech from the television averaged 61 dBA at the listeners ear,
which was usually approximately 3 m from its
loudspeaker.
At conversation distances in hospital
settings, speech averaged 55 dBA in patients
rooms and 57 dBA in the nurses stations
(Table 3). Noise levels were 45 dBA in the
former and 52 dBA in the latter. Apparently,
the talkers adjusted their speech levels only
minimally when moving from the patients
rooms to the nurses stations, despite the 7 dB
higher noise level at the nurses stations. The
S/N ratio was +10 dB in the patients rooms and
+5 dB in the nurses stations at conversation
distances.
TABLE 2. Mean speech and background levels in dBA, indoors and outdoors,
for urban and suburban homes. Standard deviations in ( ).

In the department stores, the average

background noise was slightly greater, 54 dBA.
Normalized to 1 m, the speech level was 58
dBA, but the communication distance was
somewhat closer in that setting, resulting in an
overall level of 61 dBA at the listeners ear.
The S/N ratio was approximately +7 dB at
usual conversation distances.
Noise levels were considerably higher in the
trains and airplanes, 74 and 79 dBA, respectively (Table 4). The conversation distance in
these situations was 0.4 m, and the average
speech levels at this distance were 73 dBA in
the trains and 77 dBA in the airplanes. Hence,
the S/N ratios were -1 and -2 dB in these
situations.

Anechoic Chamber Measurements

Method
Speech levels were measured in an anechoic
chamber for 100 individuals speaking at levels
ranging from casual conversation to shouting at
a distance of 1 m. The phrase, Joe took
fathers shoe bench out; she was waiting at my
lawn was uttered by females, males, and
children. They were instructed to repeat the
desired phrase at normal, raised, loud,
and shouted levels. In addition, recordings
were made of casual conversation at a distance
of 1 m in the anechoic chamber.
TABLE 4. Mean speech and background levels in
dBA in trains and airplanes. Standard deviations
in ( ).

Speech
Background Noise

Outside
Inside

1m

Speech
Conversation Distance

Urban

Suburban

Urban

Suburban

Urban

Suburban

61 (5
48 (2)

48 (4)
41 (3)

65 (4)
57 (6)

55 (5)
55 (5)

66 (4)
57 (6)

5 (5)
55 (5)

Note. All values are rounded to the nearest dB. From Table II in Pearsons et al.
(1977).

American Journal of Audiology Vol. 7 1059-0889

Background
Conversation
Noise
1m
Distance
Train
Airplanes

74 (3)
79 (3)

66 (2)
68 (4)

73 (3)
77 (4)

Note. All values are rounded to the nearest dB. From

Table II in Pearsons et al. (1977).

October 1998

Again, the recordings were analyzed with a

real-time one-third octave analysis system.
Both A-weighted and overall sound pressure
levels were determined. For the one-third
octave analysis, 100 samples at 0.1 s intervals
were taken. The integration time was equal to
fast on sound-level meters.

TABLE 5. Mean speech levels in dBA and unweighted sound pressure levels
for casual, normal, raised, loud, and shouted speech by males, females, and
children in an anechoic chamber. Unweighted sound pressure levels in [ ].
Standard deviations in ( ).

Females
Males
Children

Results
A summary of the speech levels observed for
females and males aged 13 to 60 years and for
children under age 13 for different vocal efforts
is provided in Table 5. Overall (unweighted)
and A-weighted sound pressure levels and
standard deviations are included. The differences between A-weighted and unweighted
sound pressure levels were 3 dB to 4 dB for
casual speech and for normal speech, 2 dB to 3
dB for raised speech, 1 dB for loud speech, and
0 dB for shouted speech for the three groups of
talkers.
Casual speech during conversation was
approximately 5 dB weaker than normal vocal
effort for the phrase Joe took fathers shoe
bench out; she was waiting at my lawn. Raised
speech was approximately 7 dB more intense
than normal speech, and loud speech was
elevated by another 11 dB by the males and 8
dB to 9 dB by the females and children.

Casual

Normal

Raised

Loud

Shouted

50 [54] (4)
52 [56] (4)
53 [56] (5)

55 [58] (4)
58 [61] (4)
58 [61] (5)

63 [65] (4)
65 [68] (5)
65 [67] (7)

71 [72] (6)
76 [77] (6)
74 [75] (9)

82 [82] (7)
89 [89] (7)
82 [82] (9)

Note. All values are rounded to the nearest dB. From Table I and Figures 16, 17,
and 18 in Pearsons et al. (1977).

Shouted speech increased by an additional 8 dB

to 13 dB across the three groups.
Table 6 and Figure 1 (a through e) show the
one-third octave speech spectra for the three
groups of talkers speaking at levels ranging
from casual to shouted in the anechoic chamber. The data in Table 6 are arranged by talkers
for comparison of levels as vocal effort increased from casual to shouted for each group.
The plots in Figure 1 are organized by vocal
effort for comparison of speech spectra across
talkers at each vocal effort.
The speech spectrum for the males revealed
more energy at 125 Hz, but from 250 Hz
through 8000 Hz, the speech spectra were
similar for the three groups for casual

TABLE 6. Mean sound pressure levels in one-third octave bands for speech at five vocal efforts by three groups of talkers.
Frequency
Vocal Effort 125

160

200

250

315

400

500

630

800

1000 1250 1600 2000 2500 3150 4000

5000 6300 8000

26
26
25
20
20

36
37
35
34
30

47
48
48
45
40

44
47
53
55
50

39
42
49
55
55

45
49
53
55
61

45
50
57
59
64

44
48
55
62
68

39
46
54
62
70

36
42
52
61
72

36
43
53
62
74

36
42
52
62
74

35
38
47
57
70

31
36
45
54
67

30
38
45
53
67

30
40
45
53
64

30
36
42
49
60

32
36
40
47
57

33
34
39
45
55

44
48
51
50
42

42
43
49
55
42

46
48
51
57
57

46
52
56
60
66

44
51
55
62
68

47
53
58
65
69

48
54
60
68
74

45
52
58
69
78

38
46
54
66
78

37
45
53
65
78

39
47
54
67
80

39
44
51
64
78

35
40
47
59
73

33
41
47
59
72

33
41
46
57
70

35
38
44
55
68

34
34
40
49
62

33
35
41
48
61

32
32
38
46
59

23
24
25
24
24

25
29
30
29
45

42
42
41
41
43

47
51
51
51
49

39
47
54
54
53

43
47
53
58
61

49
53
55
60
66

48
52
58
60
66

43
49
57
63
67

39
44
54
61
72

37
43
54
62
72

38
43
54
63
72

38
41
50
60
71

33
38
46
54
65

33
39
47
54
66

33
40
48
54
66

33
38
45
52
63

33
36
43
49
57

33
37
43
48
54

Females
Casual
Normal
Raised
Loud
Shouted
Males
Casual
Normal
Raised
Loud
Shouted
Children
Casual
Normal
Raised
Loud
Shouted

Note. All values are rounded to the nearest dB. Derived from Figures 16, 17, and 18 in Pearsons et al. (1977).

Olsen

FIGURE 1. Speech spectra for females, males, and children for casual conversation (A), for normal vocal effort (B), for raised
speech (C), for loud speech (D), and for shouted speech (E) as measured by Pearsons et al. (Adapted from figures 16, 17, and 18 in
Pearsons et al., 1977).

American Journal of Audiology Vol. 7 1059-0889

October 1998

conversation and for normal and raised vocal

efforts. For loud and shouted speech, the
spectra for children and for females were
similar. Male speech spectra at these levels had
considerably more energy, at least through 1250
Hz and 1600 Hz for loud and shouted speech,
respectively. Also, a shift in the maximum onethird octave band was apparent for the loud and
shouted speech spectra, rising to 630 Hz for
loud speech and to 1250 Hz for shouted male
speech. Spectra for the females and children
showed a plateau from 800 to 1600 Hz for loud
speech and from 1000 to 1600 Hz for their
shouted speech.

train and airplanes, respectively. Their measurements in an anechoic chamber further reflected
the levels of conversational speech in a quiet
environment, as well as the levels and spectra
for different vocal efforts by females, males,
and children. Speech spectra were generally
similar for the groups of talkers for casual
conversation through raised vocal efforts. For
loud speech, and particularly for shouted
speech, male speech levels were greater than
the speech levels of the females and children.
The maximum one-third octave bands for loud
and shouted speech shifted to higher frequencies for all three groups.

Summary

Reference

The large study undertaken by Pearsons et

al. (1977) for the Environmental Protection
Agency nicely demonstrates usual speech
levels in a variety of settings in classrooms,
homes, hospitals, department stores, and
commercial transportation. In most settings,
speech levels were between 55 and 66 dBA at
conversation distances in the school, home,
hospital, and department store environments.
S/N ratios on the order of 5 to 15 dB were
maintained.
Communication distance in the trains and
airplanes was considerably less than the usual 1
m, and the speech levels were higher, 73 to 77
dBA, but still at a -1 or -2 dB S/N ratio in the

Pearsons, K. S., Bennett, R. L., & Fidell, S. (1977).

Speech levels in various noise environments
(Report No. EPA-600/1-77-025). Washington,
DC: U.S. Environmental Protection Agency.
Received August 26, 1997
Accepted October 17, 1997
First published (online) November 2, 1998
DOI: 10.1044/1059-0889(1998/012)
Contact author: Wayne O. Olsen, Section of
Audiology, Mayo Clinic, Rochester, MN 55905

Key Words: speech levels, speech spectra,

schools, homes, anechoic chamber

Olsen