Welcome to the Bengali.AI Speech recognition challenge!
\nBengali.AI is a community of researchers committed to democratizing AI research in Bangladesh.
\nIn this competition, the challenge is to detect Bengali speech in audio clips collected from diverse domains. Participant have to consider the out of distribution challenge that is common in real world ML application.
\nPlease go through the competition paper for more details about the dataset.
\nAlso, checkout the pre-trained models provided by Bengali.AI.
\nAdditionally, checkout the DL Sprint community competition which was held on a smaller subset of this data.
Looking forward to an exciting competition!
", "messages": [ { "id": 2348730, "postDate": "2023-07-17T23:07:08.743Z", "content": "Welcome to the Bengali.AI Speech recognition challenge!
\nBengali.AI is a community of researchers committed to democratizing AI research in Bangladesh.
\nIn this competition, the challenge is to detect Bengali speech in audio clips collected from diverse domains. Participant have to consider the out of distribution challenge that is common in real world ML application.
\nPlease go through the competition paper for more details about the dataset.
\nAlso, checkout the pre-trained models provided by Bengali.AI.
\nAdditionally, checkout the DL Sprint community competition which was held on a smaller subset of this data.
Looking forward to an exciting competition!
", "rawMarkdown": "Welcome to the Bengali.AI Speech recognition challenge!\n\n[Bengali.AI](https://www.facebook.com/groups/bengaliAI/) is a community of researchers committed to democratizing AI research in Bangladesh. \n\nIn this competition, the challenge is to detect Bengali speech in audio clips collected from diverse domains. Participant have to consider the out of distribution challenge that is common in real world ML application.\n\nPlease go through the competition [paper](https://arxiv.org/abs/2305.09688) for more details about the dataset.\nAlso, checkout the [pre-trained models](https://huggingface.co/bengaliAI/BanglaConformer) provided by Bengali.AI.\nAdditionally, checkout the [DL Sprint](https://www.kaggle.com/competitions/dlsprint) community competition which was held on a smaller subset of this data.\n\nLooking forward to an exciting competition!\n\n\n", "votes": 20 }, { "id": 2348802, "postDate": "2023-07-18T01:50:19.693Z", "content": "It's a pleasure to participated on this Bengali AI.
\nBy the way, it's my second participation, since on DL Sprint - BUET CSE Fest 2022, I contributed with \"Levenshtein Distance Attempt\".
It's a pleasure to have you too :)
", "rawMarkdown": "It's a pleasure to have you too :)", "votes": 1 } ] }, { "id": 2351285, "postDate": "2023-07-20T02:58:07.210Z", "content": "It seems a really interesting challenge ! I am really excited as a Bengali to be able to participate in this competition! I hope to learn a lot from this competition.
", "rawMarkdown": "It seems a really interesting challenge ! I am really excited as a Bengali to be able to participate in this competition! I hope to learn a lot from this competition.", "votes": 3 }, { "id": 2349834, "postDate": "2023-07-18T18:06:04.530Z", "content": "Thank you! I am excited to work with audio for the first time!
", "rawMarkdown": "Thank you! I am excited to work with audio for the first time!", "votes": 3 }, { "id": 2348820, "postDate": "2023-07-18T02:31:03.777Z", "content": "As a Bengali myself, this challenge will be interesting :D
", "rawMarkdown": "As a Bengali myself, this challenge will be interesting :D", "votes": 3 }, { "id": 2415376, "postDate": "2023-08-30T09:15:18.683Z", "content": "I can't even see the fonts in my pc, what font is in the dataset?
", "rawMarkdown": "I can't even see the fonts in my pc, what font is in the dataset?", "votes": 1, "replies": [ { "id": 2415575, "postDate": "2023-08-30T13:02:53.270Z", "content": "Hi Ameer,
\nTry kalpurush.ttf
may I ask if the font is available on Ubuntu?
", "rawMarkdown": "may I ask if the font is available on Ubuntu?" } ] } ] }, { "id": 2374117, "postDate": "2023-08-04T17:34:03.170Z", "content": "my first recognition competition!!! I'll do my best!!
", "rawMarkdown": "my first recognition competition!!! I'll do my best!!", "votes": 1 }, { "id": 2364106, "postDate": "2023-07-29T06:06:23.397Z", "content": "Very excited. The challenge seems very interesting.
", "rawMarkdown": "Very excited. The challenge seems very interesting.", "votes": 1 }, { "id": 2363001, "postDate": "2023-07-28T12:40:47.453Z", "content": "Is this where the Bengali Speakers Unites? :P
\nAnyways, very excited to take part in the Contest.!
", "rawMarkdown": "Is this where the Bengali Speakers Unites? :P\n\nAnyways, very excited to take part in the Contest.!", "votes": 1 }, { "id": 2354394, "postDate": "2023-07-22T11:53:39.053Z", "content": "This is a good start, right？
", "rawMarkdown": "This is a good start, right？", "votes": 1 }, { "id": 2353007, "postDate": "2023-07-21T11:49:44.047Z", "content": "这是我第一次参加的kaggle比赛
", "rawMarkdown": "这是我第一次参加的kaggle比赛", "votes": 1, "replies": [ { "id": 2415570, "postDate": "2023-08-30T12:56:35.933Z", "content": "祝你好运 XuJinyi2003
", "rawMarkdown": "祝你好运 XuJinyi2003" } ] }, { "id": 2351565, "postDate": "2023-07-20T08:26:36.180Z", "content": "I don't know how to start
", "rawMarkdown": "I don't know how to start", "votes": 1 }, { "id": 2352013, "postDate": "2023-07-20T14:53:43.053Z", "content": "Excited to be a part of this competition. This is my first Kaggle competition. Looking forward to learning more about audio data and building learning models on the same in an attempt to solve this interesting and challenging ML problem.
", "rawMarkdown": "Excited to be a part of this competition. This is my first Kaggle competition. Looking forward to learning more about audio data and building learning models on the same in an attempt to solve this interesting and challenging ML problem. ", "votes": 2 }, { "id": 2353071, "postDate": "2023-07-21T12:24:30.943Z", "content": "Can you clarify the annotation generation of the train split, are they puerly generated by computers?
", "rawMarkdown": "Can you clarify the annotation generation of the train split, are they puerly generated by computers?", "replies": [ { "id": 2377476, "postDate": "2023-08-07T07:02:50.120Z", "content": "Nope the labels are not computer generated. The train split contains scripted speech, meaning contributors were given a piece of text that they had to read out into a microphone. Since the train split was crowdsourced, there is supposed to be label noise in the data that was partially removed through automated methods. We'll be sharing some methods for further cleaning/subsampling the training data soon!
", "rawMarkdown": "Nope the labels are not computer generated. The train split contains scripted speech, meaning contributors were given a piece of text that they had to read out into a microphone. Since the train split was crowdsourced, there is supposed to be label noise in the data that was partially removed through automated methods. We'll be sharing some methods for further cleaning/subsampling the training data soon!" } ] }, { "id": 2415618, "postDate": "2023-08-30T13:59:06.273Z", "rawMarkdown": "", "isDeleted": true }, { "id": 2400998, "postDate": "2023-08-21T11:19:56.170Z", "rawMarkdown": "", "isDeleted": true }, { "id": 2375510, "postDate": "2023-08-05T17:27:22.270Z", "rawMarkdown": "", "isDeleted": true }, { "id": 2349295, "postDate": "2023-07-18T09:39:55.503Z", "rawMarkdown": "", "isDeleted": true } ], "comments": [ { "id": 2348802, "author_name": "Marília Prata", "author_url": "", "post_date": "2023-07-18T01:50:19.693000", "content": "It's a pleasure to participated on this Bengali AI.
\nBy the way, it's my second participation, since on DL Sprint - BUET CSE Fest 2022, I contributed with \"Levenshtein Distance Attempt\".
It's a pleasure to have you too :)
", "votes": 1, "replies": [] } ] }, { "id": 2351285, "author_name": "Mahmud Hasan", "author_url": "", "post_date": "2023-07-20T02:58:07.210000", "content": "It seems a really interesting challenge ! I am really excited as a Bengali to be able to participate in this competition! I hope to learn a lot from this competition.
", "votes": 3, "replies": [] }, { "id": 2349834, "author_name": "Cody_Null", "author_url": "", "post_date": "2023-07-18T18:06:04.530000", "content": "Thank you! I am excited to work with audio for the first time!
", "votes": 3, "replies": [] }, { "id": 2348820, "author_name": "Rounak Kumbhakar", "author_url": "", "post_date": "2023-07-18T02:31:03.777000", "content": "As a Bengali myself, this challenge will be interesting :D
", "votes": 3, "replies": [] }, { "id": 2415376, "author_name": "Ameer Ul Islam", "author_url": "", "post_date": "2023-08-30T09:15:18.683000", "content": "I can't even see the fonts in my pc, what font is in the dataset?
", "votes": 1, "replies": [ { "id": 2415575, "author_name": "Marília Prata", "author_url": "", "post_date": "2023-08-30T13:02:53.270000", "content": "Hi Ameer,
\nTry kalpurush.ttf
may I ask if the font is available on Ubuntu?
", "votes": 0, "replies": [] } ] } ] }, { "id": 2374117, "author_name": "Shake is all I need", "author_url": "", "post_date": "2023-08-04T17:34:03.170000", "content": "my first recognition competition!!! I'll do my best!!
", "votes": 1, "replies": [] }, { "id": 2364106, "author_name": "Muhammad Sudipto Siam Dip", "author_url": "", "post_date": "2023-07-29T06:06:23.397000", "content": "Very excited. The challenge seems very interesting.
", "votes": 1, "replies": [] }, { "id": 2363001, "author_name": "PagalLadka", "author_url": "", "post_date": "2023-07-28T12:40:47.453000", "content": "Is this where the Bengali Speakers Unites? :P
\nAnyways, very excited to take part in the Contest.!
", "votes": 1, "replies": [] }, { "id": 2354394, "author_name": "KINGACE", "author_url": "", "post_date": "2023-07-22T11:53:39.053000", "content": "This is a good start, right？
", "votes": 1, "replies": [] }, { "id": 2353007, "author_name": "XuJinyi2003", "author_url": "", "post_date": "2023-07-21T11:49:44.047000", "content": "这是我第一次参加的kaggle比赛
", "votes": 1, "replies": [ { "id": 2415570, "author_name": "Marília Prata", "author_url": "", "post_date": "2023-08-30T12:56:35.933000", "content": "祝你好运 XuJinyi2003
", "votes": 0, "replies": [] } ] }, { "id": 2351565, "author_name": "Bangla News Better Pictures খবর আছে KHOBOR ACHE", "author_url": "", "post_date": "2023-07-20T08:26:36.180000", "content": "I don't know how to start
", "votes": 1, "replies": [] }, { "id": 2352013, "author_name": "Shreyak Chakraborty", "author_url": "", "post_date": "2023-07-20T14:53:43.053000", "content": "Excited to be a part of this competition. This is my first Kaggle competition. Looking forward to learning more about audio data and building learning models on the same in an attempt to solve this interesting and challenging ML problem.
", "votes": 2, "replies": [] }, { "id": 2353071, "author_name": "Aphysict", "author_url": "", "post_date": "2023-07-21T12:24:30.943000", "content": "Can you clarify the annotation generation of the train split, are they puerly generated by computers?
", "votes": 0, "replies": [ { "id": 2377476, "author_name": "Ahmed Imtiaz Humayun", "author_url": "", "post_date": "2023-08-07T07:02:50.120000", "content": "Nope the labels are not computer generated. The train split contains scripted speech, meaning contributors were given a piece of text that they had to read out into a microphone. Since the train split was crowdsourced, there is supposed to be label noise in the data that was partially removed through automated methods. We'll be sharing some methods for further cleaning/subsampling the training data soon!
", "votes": 0, "replies": [] } ] }, { "id": 2415618, "author_name": "", "author_url": "", "post_date": "2023-08-30T13:59:06.273000", "content": "", "votes": 0, "replies": [] }, { "id": 2400998, "author_name": "", "author_url": "", "post_date": "2023-08-21T11:19:56.170000", "content": "", "votes": 0, "replies": [] }, { "id": 2375510, "author_name": "", "author_url": "", "post_date": "2023-08-05T17:27:22.270000", "content": "", "votes": 0, "replies": [] }, { "id": 2349295, "author_name": "", "author_url": "", "post_date": "2023-07-18T09:39:55.503000", "content": "", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "2348730": "Welcome to the Bengali.AI Speech recognition challenge!\n\n[Bengali.AI](https://www.facebook.com/groups/bengaliAI/) is a community of researchers committed to democratizing AI research in Bangladesh. \n\nIn this competition, the challenge is to detect Bengali speech in audio clips collected from diverse domains. Participant have to consider the out of distribution challenge that is common in real world ML application.\n\nPlease go through the competition [paper](https://arxiv.org/abs/2305.09688) for more details about the dataset.\nAlso, checkout the [pre-trained models](https://huggingface.co/bengaliAI/BanglaConformer) provided by Bengali.AI.\nAdditionally, checkout the [DL Sprint](https://www.kaggle.com/competitions/dlsprint) community competition which was held on a smaller subset of this data.\n\nLooking forward to an exciting competition!\n\n\n", "2348802": "It's a pleasure to participated on this Bengali AI.\nBy the way, it's my second participation, since on DL Sprint - BUET CSE Fest 2022, I contributed with \"Levenshtein Distance Attempt\".", "2351285": "It seems a really interesting challenge ! I am really excited as a Bengali to be able to participate in this competition! I hope to learn a lot from this competition.", "2349834": "Thank you! I am excited to work with audio for the first time!", "2348820": "As a Bengali myself, this challenge will be interesting :D", "2415376": "I can't even see the fonts in my pc, what font is in the dataset?", "2374117": "my first recognition competition!!! I'll do my best!!", "2364106": "Very excited. The challenge seems very interesting.", "2363001": "Is this where the Bengali Speakers Unites? :P\n\nAnyways, very excited to take part in the Contest.!", "2354394": "This is a good start, right？", "2353007": "这是我第一次参加的kaggle比赛", "2351565": "I don't know how to start", "2352013": "Excited to be a part of this competition. This is my first Kaggle competition. Looking forward to learning more about audio data and building learning models on the same in an attempt to solve this interesting and challenging ML problem. ", "2353071": "Can you clarify the annotation generation of the train split, are they puerly generated by computers?", "2415618": "", "2400998": "", "2375510": "", "2349295": "" } }