add instructions

README.md

@@ -1,4 +1,36 @@
 # UCLA Phonetic Corpus
 
-This will contains the dataset described in the ICASSP 2021 paper
-**MULTILINGUAL PHONETIC DATASET FOR LOW RESOURCE SPEECH RECOGNITION**
+This repository contains instructions of the dataset described in our ICASSP 2021 paper `MULTILINGUAL PHONETIC DATASET FOR LOW RESOURCE SPEECH RECOGNITION`.
+
+
+We would also distribute scripts and baselines here in the future.
+
+
+If you have any suggestions or find any mistakes in the dataset, please feel free to send email to us (xinjianl [at] cs.cmu.edu) or submit an issue in this repo. Thanks!
+
+
+## Instructions
+
+Since the entire dataset is too large to be uploaded to Github, we only provide a sample of the first language (`abk`) in this repository. The full dataset can be downloaded [here](https://www.pyspeech.com/static/data/ucla_phonetic_corpus.tar.gz). 
+
+
+It is a cleaned version of the dataset in the paper. Each directory on the top level is corresponding to a language name identified by its 3 character ISO id. There are currently 97 languages in this dataset.
+
+
+Inside each directory, there will be 1 file and 1 directory
+
+- `text`: it contains the narrow phone annotations of each utterance. The first field is the utterance id.
+- `audio`: it contains all the wav format audios of each utterance. Its name is the corresponding utterance id.
+
+
+## Acknowledgements
+
+This dataset is derived from the [UCLA Phonetics Lab Archive](http://archive.phonetics.ucla.edu/). The website contains much more data and resources than we could clean in this dataset. Thank you UCLA Phonetics Lab Archive!
+
+## Reference
+
+If you find this work helpful, please cite the following paper
+
+```
+Li, Xinjian, et al. "Multilingual phonetic dataset for low resource speech recognition." ICASSP 2021-2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 2021.
+```

sample/abk/text

@@ -0,0 +1,54 @@
+abk-002-000 aˑdʒʃʲ
+abk-002-001 ˈaˑdʒmɜ
+abk-002-006 adʒɘmʃɘ́
+abk-002-009 atʃʰɜrä́ˆˑ
+abk-002-010 átʃə̆pʰɜ̆rʌ̈
+abk-002-011 áttʃʃʰɜrɜ
+abk-002-023 akʼáʒʲərɜ
+abk-002-024 ăbᵊʒʲɨ́
+abk-002-026 aˈʃæ̈́
+abk-002-027 ájəʃʲɛ̈ˇ
+abk-002-028 ˆaʃæ̈
+abk-002-030 aˆʃɘpɘ́
+abk-002-032 adʒɘ́r
+abk-002-033 adʒɘ́ʃ
+abk-002-034 adʒ
+abk-002-035 atʃədæ̈́ˇ
+abk-002-036 atʃʰnɘ́
+abk-002-037 atʃʰɘ́ɥrɜ
+abk-002-038 dɜtʃä́
+abk-002-039 atʃʰbɘ́ɡə
+abk-002-040 aptʃráˑ
+abk-002-041 atʃʼɘ́
+abk-002-042 atʃʼɘ́χrɜ
+abk-002-043 amᵊtʃʼɘ́
+abk-002-044 atʃʼá
+abk-002-045 ˈˀäʒəħʷərə
+abk-002-046 äʒᵊɹə
+abk-002-047 äʒəħœ̈ɾə
+abk-002-049 ˈˀáʒə
+abk-002-050 ˈabᵊʒə
+abk-002-051 aʃəɾɜ
+abk-002-052 áˑʃə
+abk-002-053 adʒɘ́ʃ
+abk-002-067 ˈäʁdərɜ
+abk-002-070 ˀaχɤ̈́
+abk-002-071 χpʰæ̈
+abk-002-072 ˈäχᵊrɛ̈
+abk-002-073 aχᵊrdzɛ̈
+abk-002-074 aiˇχæ̈́
+abk-002-077 ˈäχᵊrɛ̈
+abk-002-078 aχᵊrdzɛ̈
+abk-002-079 aiˇχæ̈́
+abk-002-080 amʒɤ̈́
+abk-002-083 ˈaχʲtʰɛ̈
+abk-002-084 ˈaχʲɾɛ̈
+abk-002-085 aχʲɘ́ts
+abk-002-090 atsᵊʁʷərə
+abk-002-097 aχɘ́
+abk-002-098 aχáɡə
+abk-002-101 ˀaχáˑ
+abk-002-102 aχəra
+abk-002-103 aχʷɘ́
+abk-002-105 adχʷa
+abk-002-106 anχʷa