Language endangerment poses a major challenge to linguistic diversity worldwide, and technological advances have opened new avenues for documentation and revitalization. Among these, automatic speech recognition (ASR) has shown increasing potential to assist in the transcription of endangered language data. This study focuses on Ikema, a severely endangered Ryukyuan language spoken in Okinawa, Japan, with approximately 1,300 remaining speakers, most of whom are over 60 years old. We present an ongoing effort to develop an ASR system for Ikema based on field recordings. Specifically, we (1) construct a 6.33-hour speech corpus from field recordings, (2) train an ASR model that achieves a character error rate as low as 15%, and (3) evaluate the impact of ASR assistance on the efficiency of speech transcription. Our results demonstrate that ASR integration can substantially reduce transcription time and cognitive load, offering a practical pathway toward scalable, technology-supported documentation of endangered languages.
Figures & tables
Figure 2: Map of the Miyako Islands (bottom) in Japan (top). The Ikema-speaking villages are marked with a red circle.
Dataset
#Samples
Duration
Avg. duration
(sec)
(sec)
Field
11126
16058
1.44
Dictionary
5680
5379
0.95
Audiobooks
285
1348
4.73
Total
17091
22785
1.33
Table 1: Statistics of the datasets.
Figure 3: Segmentation and annotation in ELAN. The top transcription tier contains pause-based segments, while the bottom tier contains longer segments that concatenate multiple pause-based segments to approximate sentence-level units.
kana
romaji
Model
CER
WER
CER
WER
xls-r-300m
20.94
62.56
14.80
64.99
xls-r-1b
24.07
66.61
20.33
76.73
mms-1b
21.86
63.22
18.98
73.87
Table 2: ASR performances of the trained models.
Figure 4: The curves of the CERs on the validation data during the training.
w/o ASR
w/ ASR
Speedup
Annotator A
11:36
9:43
+19.38%
Annotator B
30:49
25:00
+23.27%
Table 3: A comparison of annotation speed, with and without ASR output. The speedup is computed by dividing the time without ASR by the time with ASR.
Recording ID
Style
Duration
#Samples
#Kana
#Words
I0482_usI
Spontaneous
111.03
69
1105
202
I0482_ngi
Spontaneous
100.36
72
885
201
I0482_pic_mucIusa
Spontaneous
85.93
56
683
132
I0482_pic_buugii
Spontaneous
72.91
49
534
117
I0482_byuuigassa
Spontaneous
53.51
35
454
98
I0482_bippii
Spontaneous
26.27
15
229
43
Table 4: The detail of the spontaneous speech dataset (Field). The Audiobook data and the Dictionary data are not shown in this table.
Audio transcription is a critical bottleneck in language documentation. While multilingual Automatic Speech Recognition (ASR) models like Whisper offer solutions, field linguists often lack the expertise to utilise them. We present Easper, an open-source, no-code workflow enabling linguists to iteratively fine-tune ASR models via cloud resources directly from ELAN annotations. Deploying ASR also raises a cold start problem: deciding which recordings to transcribe first to bootstrap an accurate model. Using Easper, we evaluate transcription prioritisation strategies on three Vanuatu languages (Bislama, Nafsan, Nguna). We fine-tune models by recording session, comparing Character Error Rate trajectories when prioritising acoustic cleanliness versus linguistic richness. We demonstrate that prioritising lexically rich narratives and increasing acoustic-phonetic repetition, even in noisy environments, leads to faster improvements in transcription quality.
Aso Mahmudi, Ting Dang, Ekaterina Vylomova +1
School of Computing and Information Systems, The University of Melbourne, Australia · School of Languages and Linguistics, The University of Melbourne, Australia
Automatic Speech Recognition (ASR) technologies have achieved remarkable performance in recent years through the use of large multilingual foundation models. However, most advances remain concentrated on high-resource languages, while indigenous languages continue to suffer from a lack of speech resources and language technologies. This work presents a preliminary study on the adaptation of Whisper for Automatic Speech Recognition in Baniwa, an indigenous Arawakan language spoken in Brazil, Colombia, and Venezuela. The experiments were conducted using a corpus of 1,373 manually transcribed recordings obtained from a linguistic documentation project. The corpus contains approximately 0.54 hours of speech and consists primarily of isolated words and short elicited utterances. The Whisper Small model was fine-tuned using supervised learning and evaluated using Word Error Rate (WER) and Character Error Rate (CER). The best model achieved a WER of 37.5% and a CER of 7.45%, demonstrating that multilingual foundation models can be successfully adapted to extremely low-resource indigenous languages. The results establish an initial baseline for Baniwa Automatic Speech Recognition and provide a foundation for future research involving larger datasets, language-specific adaptation strategies, and post-processing techniques.
Leonardo Duart, Tiago Fonseca, Thiago Chacón
Department of Statistics, University of Brasilia, Brasilia, Brazil.
This study reports the development of an Automatic Speech Recognition (ASR) system in Mizo, a low-resource language. The development included collecting 17.62 hours of speech data, curating it, and fine-tuning the Mizo ASR system with three Whisper multilingual models and with the SraVaani 1.0 Indic multilingual model. Whisper-large-v3 achieved the lowest conventional WER (18.08%), while morphology-aware evaluation yielded a WER of 7.22%. Zero-shot evaluation of the SraVaani 1.0 Indic multilingual model yielded a WER of 58.27%, while Mizo-specific fine-tuning reduced the conventional WER to 29.45% and the morphology-aware WER to 17.93%. The results demonstrate that the Whisper model can achieve a substantially low WER, even when adapted to an unseen language. In contrast, SraVaani 1.0 supports the Mizo language in its multilingual model; however, fine-tuning with carefully curated Mizo speech data substantially improves its performance.