Add the v1.3

README.md

@@ -1,5 +1,20 @@
 # SALT (Sunbird African Language Technology) dataset
-SALT is a multi-way parallel text corpus of Engish and five key Ugandan languages: Luganda, Lugbara, Acholi, Runyankole and Ateso. The dataset contains 25,000 sentences covering a range of topics of local relevance, such as agriculture, health and society. Each sentence is translated into all six languages.
+
+SALT is a multi-way parallel text corpus of English and five key Ugandan languages: Luganda, Lugbara, Acholi, Runyankole and Ateso. This dataset is designed to facilitate research and development in Machine Translation (both text and speech), Automatic Speech Recognition (ASR) and Text-to-Speech synthesis (TTS). It contains sentences in both text and audio format, with audio available in both crowdsourced and studio-quality speech (work in progress).
+
+Each row in the dataset represents a sentence. This sentence is provided in six different languages (including English), with each language variant being presented in both text and audio format. Here's a breakdown of the row structure:
+
+* **text**: This section contains the written form of the sentence in six languages. Each language is represented by a key-value pair, with the key being the language code and the value being the sentence in that language.
+
+* **asr-speech**: This section contains links to the crowdsourced audio files for the sentence in six languages. Like in the text section, each language is represented by a key-value pair, where the key is the language code and the value is the filename of the audio file in that language.
+
+* **tts-speech**: This section contains links to the high-quality studio-recorded audio files for the sentence. In the current version of the dataset, only Luganda and English are available in this format.
+
+The dataset covers a wide range of locally relevant topics such as agriculture, health, and society, making it a versatile resource for developing and testing language technology applications for these languages. 
+
+Each sentence in the dataset is available as an audio file that can be accessed by appending `https://salt-tts-data.s3.eu-west-1.amazonaws.com/` to the relative paths provided in the "asr-speech" column of the dataset. This allows for the practical usage of the dataset in various speech recognition tasks. Alternatively the zip files containing each language "asr-speech" column data can be obtained from 
+
+This dataset is suited for those with a lower intermediate level of technical expertise, and can be used for a multitude of tasks including Machine Translation (text and speech), ASR, and TTS.
 
 More details are in the following publication:
 
@@ -9,13 +24,14 @@ This repository contains a notebook for training and evaluation which can be use
 
 ## Changelog
 
+#### [v1.3] 2023-07-31
+Added Multispeaker STT data for Acholi, Ateso, Luganda, Lugbara and Runyankole
+
 #### [v1.2] 2023-04-21
 Added TTS data for Luganda and English.
 
 #### [v1.1] 2023-04-19
-
 Data format changed, some duplicate sentences removed, and sentences partitioned into splits (train: 23,497; dev: 500, test: 500).
 
 #### [v1.0] 2021-12-01
-
 Initial commit of 25,000 sentences, with each sentence translated from English to Acholi, Ateso, Luganda, Lugbara and Runyankole.