Emotional Voice Database in Latvian

Voice Samples

Ten professional actors recorded nine voice samples in Latvian:

  • Four words:
    up [ˈauɡʃaː], down [ˈlejjaː], left [ˈkreisaː], right [ˈlabaː].
  • Four phrases:
    “Now press up” [ˈtaɡat ˈjaːspieʒ ˈauɡʃaː],
    “Now press down” [ˈtaɡat ˈjaːspieʒ ˈlejjaː],
    “Now press left” [ˈtaɡat ˈjaːspieʒ ˈkreisaː],
    “Now press right” [ˈtaɡat ˈjaːspieʒ ˈlabaː].
  • One paragraph:
    Four sentences from Aesop’s fable The North Wind and the Sun (Latvian version). The total length of the paragraph is 64 syllables:
    [ˈziɛ.mɛ.lis un ˈsaʊ.lɛ ˈreɪs ˈstriː.dɛː.jɑs kuɾʃ ˈsti.prɑks ˈviɲi ˈnɔs.priɛ.dɑ kɑ ˈuz.vɑ.ɾɛ.taɪs ˈbuːs ˈtas kuɾʃ ˈtsɛ.ʎi.niɛ.kɑm ˈliks ˈiz.ɟɛːɾb.tiɛs ˈvis.pirms ˈspa.ɾi.ɟi ˈuz.puː.tɑ ˈziɛ.mɛ.lis ˈtsil.vɛks ˈiɛ.ti.nɑs ˈsa.vɑs ˈdɾɛː.bɛs un ˈziɛ.mɛ.lis ˈpuː.tɑ ˈvɛːl ˈduː.ʃi.ɟɑk].

Descriptions of Emotions

Each linguistic unit was pronounced in neutral, happy, and angry intonations. Before recording, the actors prepared for the specific emotions by reading detailed descriptions of emotional states:

  • Neutral Emotion:
    Imagine that you are not experiencing strong emotions or specific interests and needs. Your task is to repeat the words, phrases, or text in a clear, neutral, and non-emotional voice.
  • Happy Emotion:
    Imagine feeling very happy and content—everything is going well, your wishes and goals are fulfilled, and circumstances are favorable. Your task is to repeat the words, phrases, or text in a clear and joyful voice.
  • Angry Emotion:
    Imagine experiencing intense anger due to obstacles or contradictions to your plans and needs. You feel frustration, resentment, and dissatisfaction. Your task is to repeat the words, phrases, or text in a clear and angry voice.

Evaluation of Voice Samples

Each voice sample was recorded three times. An expert panel of six members evaluated each sample using the following criteria:

  • Emotional Component: The extent to which the emotion is maximally expressed (anger or joy) or the voice is maximally neutral (in the case of neutral voice).
  • Voice Quality: The clarity and “purity” of the voice, including good pronunciation, diction, and articulation.

The total 810 recorded samples (10 actors × 9 linguistic units × 3 emotions × 3 repetitions) were reduced to 270 validated samples after evaluation.

Recording Conditions

The recordings were conducted in a soundproof room using the following setup:

  • Microphone: AKG C520 calibrated head-worn condenser microphone with balanced XLR audio output (AKG Acoustics, Vienna, Austria).
  • Positioning: Microphone placed laterally at a mouth-to-microphone angle of ±45° and a distance of ±5 cm.
  • Audio Interface: Scarlett Solo (Focusrite Plc., High Wycombe, United Kingdom), connected to a MacBook Air.
  • Software: PRAAT version 6.1.31 (Paul Boersma and David Weenink, Institute for Phonetic Sciences, University of Amsterdam, The Netherlands).
  • Recording Specifications: 44.1 kHz sampling rate, 16-bit resolution, and saved in .wav format.
  • Normalization: All voice samples were normalized to an intensity of 70 dB

Characteristics of Actors

The actors participating in the recordings included 5 male and 5 female speakers, aged between 23 and 35 years. Their AVQI (Acoustic Voice Quality Index) scores ranged as follows:

ActorAge (years)AVQI
F1251.82
F2232.37
F3262.06
F4232.45
F5262.55
M1272.02
M2231.99
M3241.72
M4352.37
M5242.19

Coding of Voice Recordings

The voice samples were coded using the following format:

  • M = Male, F = Female
  • 1, 2, 3, 4, 5 = Actor ID number
  • N, H, A = Emotion type (Neutral, Happy, Angry)
  • UP, DOWN, LEFT, RIGHT = Words in Latvian
  • PRESSUP, PRESSDOWN, PRESSRIGHT, PRESSLEFT = Phrases in Latvian
  • PARA = Paragraph

Example: M1_N_UP = Male actor 1, Neutral emotion, Word “Up.”

Acknowledgment

You are free to use the voice samples included in the Emotional Voice Database. Please cite the database as follows:

Trinite, B., Zdanovica, A., Kurme, D., Lavrane, E., & Magazeina, I. (2024). Emotional voice database in Latvian. Voice.liepu.lv. https://voice.liepu.lv/emotional-voice-database/

For inquiries, please contact:

F1_N_DOWN
F1_N_LEFT
F1_N_RIGHT
F1_N_UP
F1_N_PRESSDOWN
F1_N_PRESSLEFT
F1_N_PRESSRIGHT
F1_N_PRESSUP
F1_N_PARA
F2_N_DOWN
F2_N_LEFT
F2_N_RIGHT
F2_N_UP
F2_N_PRESSDOWN
F2_N_PRESSLEFT
F2_N_PRESSRIGHT
F2_N_PRESSUP
F2_N_PARA
F3_N_DOWN
F3_N_LEFT
F3_N_RIGHT
F3_N_UP
F3_N_PRESSDOWN
F3_N_PRESSLEFT
F3_N_PRESSRIGHT
F3_N_PRESSUP
F3_N_PARA
F4_N_DOWN
F4_N_LEFT
F4_N_RIGHT
F4_N_UP
F4_N_PRESSDOWN
F4_N_PRESSLEFT
F4_N_PRESSRIGHT
F4_N_PRESSUP
F4_N_PARA
F5_N_DOWN
F5_N_LEFT
F5_N_RIGHT
F5_N_UP
F5_N_PRESSDOWN
F5_N_PRESSLEFT
F5_N_PRESSRIGHT
F5_N_PRESSUP
F5_N_PARA
M1_N_DOWN
M1_N_LEFT
M1_N_RIGHT
M1_N_UP
M1_N_PRESSDOWN
M1_N_PRESSLEFT
M1_N_PRESSRIGHT
M1_N_PRESSUP
M1_N_PARA
M2_N_DOWN
M2_N_LEFT
M2_N_RIGHT
M2_N_UP
M2_N_PRESSDOWN
M2_N_PRESSLEFT
M2_N_PRESSRIGHT
M2_N_PRESSUP
M2_N_PARA
M3_N_DOWN
M3_N_LEFT
M3_N_RIGHT
M3_N_UP
M3_N_PRESSDOWN
M3_N_PRESSLEFT
M3_N_PRESSRIGHT
M3_N_PRESSUP
M3_N_PARA
M4_N_DOWN
M4_N_LEFT
M4_N_RIGHT
M4_N_UP
M4_N_PRESSDOWN
M4_N_PRESSLEFT
M4_N_PRESSRIGHT
M4_N_PRESSUP
M4_N_PARA
M5_N_DOWN
M5_N_LEFT
M5_N_RIGHT
M5_N_UP
M5_N_PRESSDOWN
M5_N_PRESSLEFT
M5_N_PRESSRIGHT
M5_N_PRESSUP
M5_N_PARA
F1_A_DOWN
F1_A_LEFT
F1_A_RIGHT
F1_A_UP
F1_A_PRESSDOWN
F1_A_PRESSLEFT
F1_A_PRESSRIGHT
F1_A_PRESSUP
F1_A_PARA
F2_A_DOWN
F2_A_LEFT
F2_A_RIGHT
F2_A_UP
F2_A_PRESSDOWN
F2_A_PRESSLEFT
F2_A_PRESSRIGHT
F2_A_PRESSUP
F2_A_PARA
F3_A_DOWN
F3_A_LEFT
F3_A_RIGHT
F3_A_UP
F3_A_PRESSDOWN
F3_A_PRESSLEFT
F3_A_PRESSRIGHT
F3_A_PRESSUP
F3_A_PARA
F4_A_DOWN
F4_A_LEFT
F4_A_RIGHT
F4_A_UP
F4_A_PRESSDOWN
F4_A_PRESSLEFT
F4_A_PRESSRIGHT
F4_A_PRESSUP
F4_A_PARA
F5_A_DOWN
F5_A_LEFT
F5_A_RIGHT
F5_A_UP
F5_A_PRESSDOWN
F5_A_PRESSLEFT
F5_A_PRESSRIGHT
F5_A_PRESSUP
F5_A_PARA
M1_A_DOWN
M1_A_LEFT
M1_A_RIGHT
M1_A_UP
M1_A_PRESSDOWN
M1_A_PRESSLEFT
M1_A_PRESSRIGHT
M1_A_PRESSUP
M1_A_PARA
M2_A_DOWN
M2_A_LEFT
M2_A_RIGHT
M2_A_UP
M2_A_PRESSDOWN
M2_A_PRESSLEFT
M2_A_PRESSRIGHT
M2_A_PRESSUP
M2_A_PARA
M3_A_DOWN
M3_A_LEFT
M3_A_RIGHT
M3_A_UP
M3_A_PRESSDOWN
M3_A_PRESSLEFT
M3_A_PRESSRIGHT
M3_A_PRESSUP
M3_A_PARA
M4_A_DOWN
M4_A_LEFT
M4_A_RIGHT
M4_A_UP
M4_A_PRESSDOWN
M4_A_PRESSLEFT
M4_A_PRESSRIGHT
M4_A_PRESSUP
M4_A_PARA
M5_A_DOWN
M5_A_LEFT
M5_A_RIGHT
M5_A_UP
M5_A_PRESSDOWN
M5_A_PRESSLEFT
M5_A_PRESSRIGHT
M5_A_PRESSUP
M5_A_PARA
F1_H_DOWN
F1_H_LEFT
F1_H_RIGHT
F1_H_UP
F1_H_PRESSDOWN
F1_H_PRESSLEFT
F1_H_PRESSRIGHT
F1_H_PRESSUP
F1_H_PARA
F2_H_DOWN
F2_H_LEFT
F2_H_RIGHT
F2_H_UP
F2_H_PRESSDOWN
F2_H_PRESSLEFT
F2_H_PRESSRIGHT
F2_H_PRESSUP
F2_H_PARA
F3_H_DOWN
F3_H_LEFT
F3_H_RIGHT
F3_H_UP
F3_H_PRESSDOWN
F3_H_PRESSLEFT
F3_H_PRESSRIGHT
F3_H_PRESSUP
F3_H_PARA
F4_H_DOWN
F4_H_LEFT
F4_H_RIGHT
F4_H_UP
F4_H_PRESSDOWN
F4_H_PRESSLEFT
F4_H_PRESSRIGHT
F4_H_PRESSUP
F4_H_PARA
F5_H_DOWN
F5_H_LEFT
F5_H_RIGHT
F5_H_UP
F5_H_PRESSDOWN
F5_H_PRESSLEFT
F5_H_PRESSRIGHT
F5_H_PRESSUP
F5_H_PARA
M1_H_DOWN
M1_H_LEFT
M1_H_RIGHT
M1_H_UP
M1_H_PRESSDOWN
M1_H_PRESSLEFT
M1_H_PRESSRIGHT
M1_H_PRESSUP
M1_H_PARA
M2_H_DOWN
M2_H_LEFT
M2_H_RIGHT
M2_H_UP
M2_H_PRESSDOWN
M2_H_PRESSLEFT
M2_H_PRESSRIGHT
M2_H_PRESSUP
M2_H_PARA
M3_H_DOWN
M3_H_LEFT
M3_H_RIGHT
M3_H_UP
M3_H_PRESSDOWN
M3_H_PRESSLEFT
M3_H_PRESSRIGHT
M3_H_PRESSUP
M3_H_PARA
M4_H_DOWN
M4_H_LEFT
M4_H_RIGHT
M4_H_UP
M4_H_PRESSDOWN
M4_H_PRESSLEFT
M4_H_PRESSRIGHT
M4_H_PRESSUP
M4_H_PARA
M5_H_DOWN
M5_H_LEFT
M5_H_RIGHT
M5_H_UP
M5_H_PRESSDOWN
M5_H_PRESSLEFT
M5_H_PRESSRIGHT
M5_H_PRESSUP
M5_H_PARA