ctsoukala commited on
Commit
41d7729
·
verified ·
1 Parent(s): fdbfcc5

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +49 -0
README.md CHANGED
@@ -18,4 +18,53 @@ configs:
18
  data_files:
19
  - split: train
20
  path: data/train-*
 
 
 
 
21
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
18
  data_files:
19
  - split: train
20
  path: data/train-*
21
+ task_categories:
22
+ - automatic-speech-recognition
23
+ tags:
24
+ - cretan
25
  ---
26
+
27
+ Cretan is a variety of Modern Greek predominantly used by
28
+ speakers who reside on the island of Crete or belong to the Cretan
29
+ diaspora. This includes communities of Cretan origin that were
30
+ relocated to the village of Hamidieh in Syria and to Western
31
+ Asia Minor, following the population exchange between Greece
32
+ and Turkey in 1923. The historical and geographical factors
33
+ that have shaped the development and preservation of the dialect
34
+ include the long-term isolation of Crete from the mainland, and
35
+ the successive domination of the island by foreign powers, such
36
+ as the Arabs, the Venetians, and the Turks, over a period of seven
37
+ centuries. Cretan has been divided based on its phonological,
38
+ phonetic, morphological, and lexical characteristics into two
39
+ major dialect groups: the western and the eastern. The boundary
40
+ between these groups coincides with the administrative division
41
+ of the island into the prefectures of Rethymno and Heraklion.
42
+ Kontosopoulos argues that the eastern dialect group is more
43
+ homogeneous than the western one, which shows more variation
44
+ across all levels of linguistic analysis. Contrary to other Modern
45
+ Greek Dialects, Cretan does not face the threat of extinction, as
46
+ it remains the sole means of communication for a large number
47
+ of speakers in various parts of the island.
48
+
49
+ For the compilation of this corpus, a linguist (Socrates Varkitzian) gathered 32 tapes
50
+ containing material from radio broadcasts in digital format, with
51
+ permission from the Audiovisual Department of the Vikelaia
52
+ Municipal Library of Heraklion, Crete. These broadcasts were
53
+ recorded and aired by Radio Mires, in the Messara region of
54
+ Heraklion, during the period 1998-2001, totaling 958 minutes
55
+ and 47 seconds. These recordings primarily consist of narratives
56
+ by one speaker, Ioannis Anagnostakis, who is responsible for
57
+ their composition. In terms of textual genre, the linguistic content of the broadcasts consists of folklore narratives expressed in
58
+ the local linguistic variety. Out of the total volume of material
59
+ collected, we utilized nine tapes. Criteria for material selection
60
+ included, on the one hand, maximizing digital clarity of speech
61
+ and, on the other hand, ensuring representative sampling across
62
+ the entire three-year period of radio recordings. To obtain an
63
+ initial transcription, we employed the Large-v2 model, which
64
+ was the largest Whisper model at the time. Subsequently, the
65
+ transcripts were manually corrected in collaboration with the
66
+ local community. The transcription system that was used was
67
+ based on the Greek alphabet and orthography.
68
+
69
+ This corpus was used in the Interspeech proceedings paper [Speech Recognition for Greek Dialects: A Challenging Benchmark](https://www.isca-archive.org/interspeech_2024/vakirtzian24_interspeech.html).
70
+ The paper presented a comprehensive study of automatic speech recognition (ASR) for low-resource Greek varieties, focusing on the unique challenges posed by dialects.