The Estonian Adolescent Speech Corpus, funded by the national program Estonian Language Technology (2011–2017), was created at the Laboratory of Language Technology, Tallinn University of Technology.

The corpus contains recordings of 309 adolescent speakers of Estonian (134 male, 175 female), aged 9–18. Recordings were made in public schools in 10 locations across Estonia using SpeechRecorder. Both channels were recorded: a desktop microphone and a second channel (headset microphone and, for 32 subjects, a laryngograph). Sessions were conducted in quiet school rooms.

The corpus includes both read and spontaneous speech, with 67 items per speaker:

Read speech

21 phonetically rich sentences
11 sentences with geographical, personal, and company names
5 time expressions
13 telephone numbers and digit strings
5 IT-related terms
2 short stories

Spontaneous speech

3 picture descriptions
7 free talks (about themselves, family, school, friends, hobbies, or freely chosen topics)

Transcriptions

Made with Transcriber
Texts also provided as UTF-8 TXT files

Corpus Structure

Each speaker's data is stored in a folder named K_xxx_g, where:
- xxx is a numeric code
- g = 1 for male speakers, g = 2 for female speakers
- Example: K_142_1 = male speaker, K_142_2 = female speaker
For 32 speakers, subfolders with an lx suffix contain signals recorded with a microphone and laryngograph
- Example: K_242_1 → K_242_1_lx

Corpus Statistics

Total recordings: 22,846
Duration: 4637 minutes (77 hours)
Size: 45.7 GB
Formats:
- WAV: 44.1 kHz, 16-bit stereo
- TRS and TXT in UTF-8 encoding

Licence

META-SHARE NonCommercial NoRedistribution Licence 2.0

The corpus is available for research and educational purposes upon request from Tallinn University of Technology.

Contact

Einar Meister
📧 [email protected]

Citing

Lya Meister, Einar Meister (2014)
Development of the Corpus of Estonian Adolescent Speech.
In: Human Language Technologies – the Baltic Perspective: Proceedings of the Sixth International Baltic Conference, Baltic HLT 2014, pp. 206–209.
Editors: Utka, Andrius; Grigonytė, Gintarė; Kapočiūtė-Dzikienė, Jurgita; Vaičenonienė, Jurgita.
Amsterdam: IOS Press. (Frontiers in Artificial Intelligence and Applications, vol. 268).

@inproceedings{Meister2014,
  author    = {Meister, Lya and Meister, Einar},
  title     = {Development of the corpus of Estonian Adolescent Speech},
  booktitle = {Human Language Technologies - the Baltic Perspective: Proceedings of the Sixth International Baltic Conference, Baltic HLT 2014},
  editor    = {Utka, Andrius and Grigonytė, Gintarė and Kapočiūtė-Dzikienė, Jurgita and Vaičenonienė, Jurgita},
  year      = {2014},
  publisher = {IOS Press},
  address   = {Amsterdam},
  pages     = {206--209},
  series    = {Frontiers in Artificial Intelligence and Applications},
  volume    = {268},
  doi       = {10.3233/978-1-61499-442-8-206}
}

Einar Meister, Lya Meister (2017)
Eesti laste kõne I. Põhitooni akustiline analüüs.
Keel ja Kirjandus, 60(7), 518–533.

@article{Meister2017,
  author    = {Meister, Einar and Meister, Lya},
  title     = {Eesti laste kõne I. Põhitooni akustiline analüüs},
  journal   = {Keel ja Kirjandus},
  year      = {2017},
  volume    = {60},
  number    = {7},
  pages     = {518--533},
  doi       = {10.54013/kk716a2}
}

Downloads last month: 6

You need to agree to share your contact information to access this dataset

Estonian Adolescent Speech Corpus 1.0

Contents