Skip to main content

Table 1 MAVIR database characteristics. “train/dev” stands for training/development, “occ.” stands for occurrences, “min” stands for minutes, “SNR” for signal-to-noise ratio, and “dB” for decibels

From: Spoken term detection ALBAYZIN 2014 evaluation: overview, systems, results, and discussion

File ID

Dataset

# word occ.

Duration (min)

# speakers

SNR

MAVIR-02

train/dev

13432

74.51

7 (7 male)

2.1 dB

MAVIR-03

train/dev

6681

38.18

2 (1 male, 1 female)

15.8 dB

MAVIR-06

train/dev

4332

29.15

3 (2 males, 1 female)

12.0 dB

MAVIR-07

train/dev

3831

21.78

2 (2 males)

10.6 dB

MAVIR-08

train/dev

3356

18.90

1 (1 male)

7.5 dB

MAVIR-09

train/dev

11179

70.05

1 (1 male)

12.3 dB

MAVIR-12

train/dev

11168

67.66

1 (1 male)

11.1 dB

MAVIR-04

test

9310

57.36

4 (3 males, 1 female)

10.2 dB

MAVIR-11

test

3130

20.33

1 (1 male)

9.2 dB

MAVIR-13

test

7837

43.61

1 (1 male)

11.1 dB

ALL

train/dev

53979

320.23

17 (15 males and 2 females)

-

ALL

test

20277

121.3

6 (5 males and 1 female)

-