{ "id": 88001, "title": "Some learning materials", "url": "/competitions/freesound-audio-tagging-2019/discussion/88001", "author_name": "", "post_date": "2019-04-05T04:51:18.091758500Z", "votes": 18, "comment_count": 3, "views": 0, "content": "

For speech recognition, traditional models like GMM/HMM are still very powerful. Stanford's book provides all the details (esp. chapter 9): https://github.com/rain1024/slp2-pdf

\n\n

Baidu's deep speech paper (Ng etc. https://arxiv.org/pdf/1412.5567.pdf) released the power of end2end deep learning in this field. Here are some Keras implementations: https://github.com/robmsmt/KerasDeepSpeech/blob/master/model.py

\n\n

For a detailed understanding of deep learning in speech, Deng Li's book is great: https://www.amazon.com/Automatic-Speech-Recognition-Communication-Technology/dp/1447157788. It is expensive. His slides may be good enough: https://www.microsoft.com/en-us/research/wp-content/uploads/2016/07/interspeech-tutorial-2015-lideng-sept6a.pdf

\n\n

For more worth-reading papers, I would like to recommend this list: https://github.com/zzw922cn/awesome-speech-recognition-speech-synthesis-papers

\n\n

Some good kernels from last time: https://www.kaggle.com/c/freesound-audio-tagging/kernels?sortBy=scoreDescending&group=everyone&pageSize=20&competitionId=8900

\n\n

Happy Kaggling!

", "messages": [ { "id": "507685", "postDate": "04/05/2019 04:51:18", "content": "

For speech recognition, traditional models like GMM/HMM are still very powerful. Stanford's book provides all the details (esp. chapter 9): https://github.com/rain1024/slp2-pdf

\n\n

Baidu's deep speech paper (Ng etc. https://arxiv.org/pdf/1412.5567.pdf) released the power of end2end deep learning in this field. Here are some Keras implementations: https://github.com/robmsmt/KerasDeepSpeech/blob/master/model.py

\n\n

For a detailed understanding of deep learning in speech, Deng Li's book is great: https://www.amazon.com/Automatic-Speech-Recognition-Communication-Technology/dp/1447157788. It is expensive. His slides may be good enough: https://www.microsoft.com/en-us/research/wp-content/uploads/2016/07/interspeech-tutorial-2015-lideng-sept6a.pdf

\n\n

For more worth-reading papers, I would like to recommend this list: https://github.com/zzw922cn/awesome-speech-recognition-speech-synthesis-papers

\n\n

Some good kernels from last time: https://www.kaggle.com/c/freesound-audio-tagging/kernels?sortBy=scoreDescending&group=everyone&pageSize=20&competitionId=8900

\n\n

Happy Kaggling!

", "rawMarkdown": "For speech recognition, traditional models like GMM/HMM are still very powerful. Stanford's book provides all the details (esp. chapter 9): https://github.com/rain1024/slp2-pdf\n\nBaidu's deep speech paper (Ng etc. https://arxiv.org/pdf/1412.5567.pdf) released the power of end2end deep learning in this field. Here are some Keras implementations: https://github.com/robmsmt/KerasDeepSpeech/blob/master/model.py \n\nFor a detailed understanding of deep learning in speech, Deng Li's book is great: https://www.amazon.com/Automatic-Speech-Recognition-Communication-Technology/dp/1447157788. It is expensive. His slides may be good enough: https://www.microsoft.com/en-us/research/wp-content/uploads/2016/07/interspeech-tutorial-2015-lideng-sept6a.pdf\n\nFor more worth-reading papers, I would like to recommend this list: https://github.com/zzw922cn/awesome-speech-recognition-speech-synthesis-papers\n\nSome good kernels from last time: https://www.kaggle.com/c/freesound-audio-tagging/kernels?sortBy=scoreDescending&group=everyone&pageSize=20&competitionId=8900\n\nHappy Kaggling!", "votes": null }, { "id": "507714", "postDate": "04/05/2019 05:31:45", "content": "

thanks for sharing

", "rawMarkdown": "thanks for sharing", "votes": null }, { "id": "508227", "postDate": "04/05/2019 20:23:59", "content": "

Thanks for sharing!

", "rawMarkdown": "Thanks for sharing!", "votes": null }, { "id": "508417", "postDate": "04/06/2019 06:12:24", "content": "

thank you very much for sharing

", "rawMarkdown": "thank you very much for sharing", "votes": null } ], "comments": [ { "id": 507714, "author_name": "lohitharcot", "author_url": "", "post_date": "04/05/2019 05:31:45", "content": "

thanks for sharing

", "votes": null, "replies": [] }, { "id": 508227, "author_name": "dhaqui", "author_url": "", "post_date": "04/05/2019 20:23:59", "content": "

Thanks for sharing!

", "votes": null, "replies": [] }, { "id": 508417, "author_name": "malkoch", "author_url": "", "post_date": "04/06/2019 06:12:24", "content": "

thank you very much for sharing

", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "507685": "For speech recognition, traditional models like GMM/HMM are still very powerful. Stanford's book provides all the details (esp. chapter 9): https://github.com/rain1024/slp2-pdf\n\nBaidu's deep speech paper (Ng etc. https://arxiv.org/pdf/1412.5567.pdf) released the power of end2end deep learning in this field. Here are some Keras implementations: https://github.com/robmsmt/KerasDeepSpeech/blob/master/model.py \n\nFor a detailed understanding of deep learning in speech, Deng Li's book is great: https://www.amazon.com/Automatic-Speech-Recognition-Communication-Technology/dp/1447157788. It is expensive. His slides may be good enough: https://www.microsoft.com/en-us/research/wp-content/uploads/2016/07/interspeech-tutorial-2015-lideng-sept6a.pdf\n\nFor more worth-reading papers, I would like to recommend this list: https://github.com/zzw922cn/awesome-speech-recognition-speech-synthesis-papers\n\nSome good kernels from last time: https://www.kaggle.com/c/freesound-audio-tagging/kernels?sortBy=scoreDescending&group=everyone&pageSize=20&competitionId=8900\n\nHappy Kaggling!", "507714": "thanks for sharing", "508227": "Thanks for sharing!", "508417": "thank you very much for sharing" }, "source": "meta" }