{ "id": 334477, "title": "Validation, test and training files in wav format ", "url": "/competitions/dlsprint/discussion/334477", "author_name": "", "post_date": "2022-07-01T18:03:46.335850200Z", "votes": 14, "comment_count": 2, "views": 0, "content": "

The audios in the dataset are given in mp3 format. Most of the ASR models require the data in wav format. I have converted the validation, test and train(partially) audios into wav format and uploaded them as datasets. Feel free to use them.

\n

Validation Dataset : https://www.kaggle.com/datasets/mbmmurad/validation-fileswav-format

\n

Test Dataset : https://www.kaggle.com/datasets/mbmmurad/test-wav-files-dl-sprint

\n

Training Dataset : https://www.kaggle.com/datasets/mbmmurad/train-mp3-40k

\n

( This dataset contains 40k samples. Converting all the 200k+ audios Isn't possible since the memory keeps running out. I'll be adding the rest of the samples gradually)

\n

Refer to this notebook to see how the files were converted.

\n

Please refer to this notebook to see a faster of converting mp3 to wav.

", "messages": [ { "id": "1839832", "postDate": "07/01/2022 18:03:46", "content": "

The audios in the dataset are given in mp3 format. Most of the ASR models require the data in wav format. I have converted the validation, test and train(partially) audios into wav format and uploaded them as datasets. Feel free to use them.

\n

Validation Dataset : https://www.kaggle.com/datasets/mbmmurad/validation-fileswav-format

\n

Test Dataset : https://www.kaggle.com/datasets/mbmmurad/test-wav-files-dl-sprint

\n

Training Dataset : https://www.kaggle.com/datasets/mbmmurad/train-mp3-40k

\n

( This dataset contains 40k samples. Converting all the 200k+ audios Isn't possible since the memory keeps running out. I'll be adding the rest of the samples gradually)

\n

Refer to this notebook to see how the files were converted.

\n

Please refer to this notebook to see a faster of converting mp3 to wav.

", "rawMarkdown": "The audios in the dataset are given in mp3 format. Most of the ASR models require the data in wav format. I have converted the validation, test and train(partially) audios into wav format and uploaded them as datasets. Feel free to use them.\n\nValidation Dataset : https://www.kaggle.com/datasets/mbmmurad/validation-fileswav-format\n\nTest Dataset : https://www.kaggle.com/datasets/mbmmurad/test-wav-files-dl-sprint\n\nTraining Dataset : https://www.kaggle.com/datasets/mbmmurad/train-mp3-40k\n\n( This dataset contains 40k samples. Converting all the 200k+ audios Isn't possible since the memory keeps running out. I'll be adding the rest of the samples gradually) \n\nRefer to this [notebook](https://www.kaggle.com/code/mbmmurad/mp3-to-wav-conversion) to see how the files were converted.\n\nPlease refer to this [notebook](https://www.kaggle.com/code/mbmmurad/faster-way-to-convert-mp3-to-wav-using-joblib) to see a faster of converting mp3 to wav.", "votes": null }, { "id": "1848312", "postDate": "07/08/2022 14:07:41", "content": "

all train audio wavs
\nhttps://www.kaggle.com/datasets/ocrteamriad/train-wavs-all-dl-sprint

", "rawMarkdown": "all train audio wavs\nhttps://www.kaggle.com/datasets/ocrteamriad/train-wavs-all-dl-sprint", "votes": null }, { "id": "1849082", "postDate": "07/09/2022 07:33:16", "content": "

https://www.kaggle.com/datasets/tasnimnishatislam/bangla-train
\ntrain wav files

", "rawMarkdown": "https://www.kaggle.com/datasets/tasnimnishatislam/bangla-train\ntrain wav files", "votes": null } ], "comments": [ { "id": 1848312, "author_name": "nazmuddhohaansary", "author_url": "", "post_date": "07/08/2022 14:07:41", "content": "

all train audio wavs
\nhttps://www.kaggle.com/datasets/ocrteamriad/train-wavs-all-dl-sprint

", "votes": null, "replies": [] }, { "id": 1849082, "author_name": "tasnimnishatislam", "author_url": "", "post_date": "07/09/2022 07:33:16", "content": "

https://www.kaggle.com/datasets/tasnimnishatislam/bangla-train
\ntrain wav files

", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "1839832": "The audios in the dataset are given in mp3 format. Most of the ASR models require the data in wav format. I have converted the validation, test and train(partially) audios into wav format and uploaded them as datasets. Feel free to use them.\n\nValidation Dataset : https://www.kaggle.com/datasets/mbmmurad/validation-fileswav-format\n\nTest Dataset : https://www.kaggle.com/datasets/mbmmurad/test-wav-files-dl-sprint\n\nTraining Dataset : https://www.kaggle.com/datasets/mbmmurad/train-mp3-40k\n\n( This dataset contains 40k samples. Converting all the 200k+ audios Isn't possible since the memory keeps running out. I'll be adding the rest of the samples gradually) \n\nRefer to this [notebook](https://www.kaggle.com/code/mbmmurad/mp3-to-wav-conversion) to see how the files were converted.\n\nPlease refer to this [notebook](https://www.kaggle.com/code/mbmmurad/faster-way-to-convert-mp3-to-wav-using-joblib) to see a faster of converting mp3 to wav.", "1848312": "all train audio wavs\nhttps://www.kaggle.com/datasets/ocrteamriad/train-wavs-all-dl-sprint", "1849082": "https://www.kaggle.com/datasets/tasnimnishatislam/bangla-train\ntrain wav files" }, "source": "meta" }