Datasets

Dataset

indic-diarbench

by sarvamai

Indic DiarBench A multilingual joint diarization and ASR benchmark for Indian languages, spanning all 22 scheduled languages of India with approximately 108 hours of natural multi-speaker audio.

CC-BY-4.0automatic-speech-recognition 1787 171K < n < 10K rows

published 25 Jul 2026

Preview

First 5 rows of Odia / test, from the Hugging Face dataset viewer.

audiorecording_idlanguageannotated_transcriptdataset_typesample_idnum_speakersnum_segmentsduration_seconds
[{"src":"https://datasets-server.huggingface.co/assets/sarvamai/indic-diarbench/--/92877bad8aab6e598167d91c6ee02aa8ca6ede09/--/Odia/test/0/audio/audio.wav?Expires=1788229088&Signature=mHE-UPNFpfCKE1ijCzh~Ec~F~ErTT9zeI2ktEXsO9xhIw4VjgXOSqe441vbVeNODQCAuGZBNHRdxdhqVfVP1N7Rq0CkotKL3T1inKRRlSKuSN20s13J…odia_nf_001Odia[{"speaker_id":"SPEAKER_01","transcript":"ସେଇ ପନ୍ଦର(15)-ଦିନ ସେମାନେ ସେବା କରନ୍ତି।","start_time":0.932,"end_time":3.36},{"speaker_id":"SPEAKER_06","transcript":"ସେଇ ହିସାବରେ ସେ ବଣମଲ୍ଲୀ ଲଗାନ୍ତି। ହେଲା? ଆ ମାନେ ଫଳ ଖାଆନ୍ତି। କାରଣ ଯେତେବେଳେ ସେ ସେଇଠି ଥିଲେ ତାଙ୍କୁ ବିଶ୍ଵାବସୁ ସେଇଆ ହିଁ ଚଢଉଥିଲେ। ବଣର ଫୁଲ ଦଉଥିଲେ। ଆପଣଙ୍…Near fieldodia_001557201.86
[{"src":"https://datasets-server.huggingface.co/assets/sarvamai/indic-diarbench/--/92877bad8aab6e598167d91c6ee02aa8ca6ede09/--/Odia/test/1/audio/audio.wav?Expires=1788229088&Signature=XOFkJBv1xTne8eqhOgOjtTsXGmPsorMZca2Um3vCZARC0dMbqbb~32BeyV~AazmncZWrAbvu8F812D5Z-R98G97kMjDoH~vzvlkEk0hj6gi4B4~oAzu…odia_nf_001Odia[{"speaker_id":"SPEAKER_04","transcript":"ହଁ ସିଏ କହିଲେ ସିଏ ତ ଚଣ୍ଡାଳୁଣୀ ଘରକୁ ଗଲା ନା? ତ ସିଏ ଜାତି ତାଙ୍କର ହଁ ତଳ ଜାତି। ସେତେବେଳେ ସେଇଥିପାଇଁ।","start_time":0.356,"end_time":9.248},{"speaker_id":"SPEAKER_07","transcript":"ହୁଁ।","start_time":3.364,"end_time":3.808},{"speaker_id":"SPEAKER_06","transcript":"ହଁ…Near fieldodia_002767186.94
[{"src":"https://datasets-server.huggingface.co/assets/sarvamai/indic-diarbench/--/92877bad8aab6e598167d91c6ee02aa8ca6ede09/--/Odia/test/2/audio/audio.wav?Expires=1788229088&Signature=iuGPO3VnhBsVdAYChXRW1KX71TDqFNCh1A7HI92zqn8QH4hCaBDqnxqol-tE7gZHr3z~P-YB2Wa~l7xmZeNvzrjWj9kFBEGUCNgcvGXQCxYpe171KMh…odia_nf_001Odia[{"speaker_id":"SPEAKER_01","transcript":"ଆଗକୁ ଯିବା। ଆଉ ଧନ୍ୟବାଦ। ଆମେ ନେକ୍ସଟ୍ (Next) ରାଉଣ୍ଡ(Round)-କୁ ଯିବା। ନେକ୍ସଟ୍ (Next) ରାଉଣ୍ଡ(Round)-କୁହେଉଛି ୱାର୍ଲ୍ଡ(World) ଟୁର୍(Tour)। ତ ମୁଁ କିଛି କ୍ଵୋଶ୍ଚିନ୍(question) କହିବି ଆଉ ତାର ଅପ୍ସନ୍ସ(options) ବି ଦେବି। ତ ସେଇ ଅପ୍ସନ୍ସ(options) ଭିତରୁ ଆପଣ ବାଛିକି କାହିଁକି କାହିଁକି …Near fieldodia_003742195.14
[{"src":"https://datasets-server.huggingface.co/assets/sarvamai/indic-diarbench/--/92877bad8aab6e598167d91c6ee02aa8ca6ede09/--/Odia/test/3/audio/audio.wav?Expires=1788229088&Signature=Du7k~6cOChg9OXIgLsWkmdcZ07sNENLJ0LwYMM2EGbaacBGh0Ofos5C~1tq4ZMVSB-sBQNHBSI-4GWdlKL4vvqNWHVuWoqDMrSKEtKOV2WgHSHWX8Cw…odia_nf_001Odia[{"speaker_id":"SPEAKER_09","transcript":"ସବୁଠାରୁ ଭଲ ଲାଗେ, ବେଶୀ ଭଲ ଲାଗେ ପଖାଳଟା ଖାଇବାକୁ।","start_time":1.028,"end_time":5.216},{"speaker_id":"SPEAKER_07","transcript":"ସବୁ","start_time":1.06,"end_time":1.408},{"speaker_id":"SPEAKER_06","transcript":"<unintelligible>","start_time":1.156,"end_time":2.…Near fieldodia_004872197.47
[{"src":"https://datasets-server.huggingface.co/assets/sarvamai/indic-diarbench/--/92877bad8aab6e598167d91c6ee02aa8ca6ede09/--/Odia/test/4/audio/audio.wav?Expires=1788229088&Signature=tohfcytX4PqFJojGfYcUhAIYmHlGRCKAOIx3PMhXDklGOdcqUMxyhojbgQSO1KzoakKCxPiUX~bZ7fPasySEn75i1CADMiODi-JuvV5TdYw3Ezra2aX…odia_nf_001Odia[{"speaker_id":"SPEAKER_06","transcript":"ହଁ, ଦେବଦାସୀ ନୃତ୍ୟ ଯୋଉଟା ଦେବଦାସୀ କୁହନ୍ତିନି ପୁରୀ(Puri)-ରେ, ସେଇଟାକୁ ମାହାରୀ ନୃତ୍ୟ କହନ୍ତି। ମୋ ଭଉଣୀ ମାହାରୀ ନୃତ୍ୟ କରେ। ତ ମୁଁ ଜାଣିଛି ସେଇଟା ସବୁ ଗୀତ ଗୋବିନ୍ଦ, ଆପଣଙ୍କ ଚନ୍ଦନରେ ଯୋଉ ଗୀତ ଆମର ବୋଲାଯାଏ ଲୋକଗୀତ ଯୋଉଟା।","start_time":0.452,"end_time":13.728},{"speaker_id":"SPEAKE…Near fieldodia_005757189.34

Cite this

CC-BY-4.0audiotextCite

Tags

task_categories:automatic-speech-recognitiontask_categories:audio-to-audiolanguage:aslanguage:bnlanguage:brxlanguage:doilanguage:gulanguage:hilanguage:knlanguage:kslanguage:koklanguage:mailanguage:mllanguage:mnilanguage:mrlanguage:nelanguage:orlanguage:palanguage:salanguage:satlanguage:sdlanguage:talanguage:telanguage:urlicense:cc-by-4.0size_categories:1K<n<10Kformat:parquetmodality:audiomodality:textlibrary:datasetslibrary:pandaslibrary:polarslibrary:mlcroissantarxiv:2607.23808region:usdiarizationindicspeechmulti-speakerannotation