Nexdata-AI commited on
Commit
a7ac83f
·
verified ·
1 Parent(s): 98acdef

Create README.md

Browse files
Files changed (1) hide show
  1. README.md +46 -0
README.md ADDED
@@ -0,0 +1,46 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ language:
3
+ - en
4
+ - de
5
+ - fr
6
+ - it
7
+ - ja
8
+ - ko
9
+ - pt
10
+ - es
11
+ - ru
12
+ - th
13
+ - vi
14
+ ---
15
+ ## Description
16
+ The INTERSPEECH 2025 MLC-SLM Challenge Dataset, curated by Datatang, is derived from fifteen proprietary conversational speech corpora. Distinguished by exceptional annotation accuracy and operational reliability, this dataset is engineered to address critical challenges in multilingual automatic speech recognition (ASR) and long-context comprehension. It meticulously replicates real-world complexities including spontaneous interruptions and speaker overlaps across 11 languages (1500 hours total duration), thereby providing robust training resources for developing world-ready ASR systems.
17
+
18
+ For more details, please refer to the link: https://www.nexdata.ai/datasets/speechrecog/1892?source=Huggingface
19
+
20
+ ## Specifications
21
+ ### Format
22
+
23
+ 16kHz, 16bit, uncompressed wav, mono channel;
24
+
25
+ ### Recording Environment
26
+
27
+ quiet indoor environment, without echo;
28
+
29
+ ### Recording content
30
+
31
+ dozens of topics are specified, and the speakers make dialogue under those topics while the recording is performed;
32
+
33
+ ### Annotation
34
+
35
+ annotating for the transcription text, speaker identification, gender;
36
+
37
+ ### Device
38
+
39
+ Android mobile phone, iPhone;
40
+
41
+ ### Language
42
+
43
+ American English/British English/Filipino English/Australian English/Indian English/French/German/Italian/Japanese/Korean/Portuguese(Europe)/Russian/Spanish(Spain)/Thai/Vietnamese.
44
+
45
+ ## Full Dataset Access
46
+ The complete dataset is available upon request. Reach out to us to learn more and submit an access request.