![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F753022%2F7ef20c5b40e5fba1343809cee431997e%2Fphoto_2024-10-25_18-14-41.jpg?generation=1729872622034809&alt=media\")
@jeevahimanshukaushik could you please clarify the model / encoder which were used to create the embeddings in the competition dataset? Thank you in advance.
", "messages": [ { "id": "3027929", "postDate": "10/25/2024 12:56:51", "content": "@jeevahimanshukaushik could you please clarify the model / encoder which were used to create the embeddings in the competition dataset? Thank you in advance.
", "rawMarkdown": "jeevahimanshukaushik could you please clarify the model / encoder which were used to create the embeddings in the competition dataset? Thank you in advance.", "votes": null }, { "id": "3027960", "postDate": "10/25/2024 13:37:58", "content": "@samvelkoch it is generated using googles HeAR api. Here is the github link for HeAR
\nThe preprocessing for isolation of cough sounds was done using the code given in the overview. For vowel sounds, the mid interval from 2 secs to 4 secs was selected
", "rawMarkdown": "samvelkoch it is generated using googles HeAR api. [Here](https://github.com/Google-Health/google-health/blob/master/health_acoustic_representations/README.md) is the github link for HeAR\n\nThe preprocessing for isolation of cough sounds was done using the code given in the overview. For vowel sounds, the mid interval from 2 secs to 4 secs was selected", "votes": null }, { "id": "3028018", "postDate": "10/25/2024 14:32:34", "content": "Thank you 🙏 HeAR embeddings change the game bets. Very interesting to dive deeper.
", "rawMarkdown": "Thank you 🙏 HeAR embeddings change the game bets. Very interesting to dive deeper.", "votes": null }, { "id": "3028118", "postDate": "10/25/2024 16:12:10", "content": "Could you clarify a bit shapes of embeddings? Why cough embeddings are so different? Are their shape relate to cough files duration?
@jeevahimanshukaushik and dataset provenance if possible please. How did you collect the data?
", "rawMarkdown": "jeevahimanshukaushik and dataset provenance if possible please. How did you collect the data?", "votes": null }, { "id": "3028164", "postDate": "10/25/2024 17:17:13", "content": "Each sample of cough sound contains cough bouts ( a sudden burst of air for short interval in coughing). The shape of each of these samples are based out of number of cough bouts captured by the model. Interesting thing is we can consider each cough bout as a seperate data point as well to play with
\nFor vowel sound only 2 sec audio is used from the vowel.wav file. So each of those have a shape of 1, 512
", "rawMarkdown": "Each sample of cough sound contains cough bouts ( a sudden burst of air for short interval in coughing). The shape of each of these samples are based out of number of cough bouts captured by the model. Interesting thing is we can consider each cough bout as a seperate data point as well to play with\n\nFor vowel sound only 2 sec audio is used from the vowel.wav file. So each of those have a shape of 1, 512", "votes": null }, { "id": "3028167", "postDate": "10/25/2024 17:18:53", "content": "The data was collected using mobile phone via webapp. In most of the cases, recording was done on center. For a few of them recording was done remotely
", "rawMarkdown": "The data was collected using mobile phone via webapp. In most of the cases, recording was done on center. For a few of them recording was done remotely", "votes": null }, { "id": "3058792", "postDate": "11/30/2024 01:45:49", "content": "Can you explain more detail about dataset source like the location (ex. Hospital name) or maybe the object of people, because I use this dataset for my college project?
", "rawMarkdown": "Can you explain more detail about dataset source like the location (ex. Hospital name) or maybe the object of people, because I use this dataset for my college project?", "votes": null } ], "comments": [ { "id": 3027960, "author_name": "jeevahimanshukaushik", "author_url": "", "post_date": "10/25/2024 13:37:58", "content": "@samvelkoch it is generated using googles HeAR api. Here is the github link for HeAR
\nThe preprocessing for isolation of cough sounds was done using the code given in the overview. For vowel sounds, the mid interval from 2 secs to 4 secs was selected
", "votes": null, "replies": [ { "id": 3028018, "author_name": "samvelkoch", "author_url": "", "post_date": "10/25/2024 14:32:34", "content": "Thank you 🙏 HeAR embeddings change the game bets. Very interesting to dive deeper.
", "votes": null, "replies": [] }, { "id": 3028118, "author_name": "ambisinistra", "author_url": "", "post_date": "10/25/2024 16:12:10", "content": "Could you clarify a bit shapes of embeddings? Why cough embeddings are so different? Are their shape relate to cough files duration?
@jeevahimanshukaushik and dataset provenance if possible please. How did you collect the data?
", "votes": null, "replies": [ { "id": 3028167, "author_name": "jeevahimanshukaushik", "author_url": "", "post_date": "10/25/2024 17:18:53", "content": "The data was collected using mobile phone via webapp. In most of the cases, recording was done on center. For a few of them recording was done remotely
", "votes": null, "replies": [ { "id": 3058792, "author_name": "husninm", "author_url": "", "post_date": "11/30/2024 01:45:49", "content": "Can you explain more detail about dataset source like the location (ex. Hospital name) or maybe the object of people, because I use this dataset for my college project?
", "votes": null, "replies": [] } ] } ] }, { "id": 3028164, "author_name": "jeevahimanshukaushik", "author_url": "", "post_date": "10/25/2024 17:17:13", "content": "Each sample of cough sound contains cough bouts ( a sudden burst of air for short interval in coughing). The shape of each of these samples are based out of number of cough bouts captured by the model. Interesting thing is we can consider each cough bout as a seperate data point as well to play with
\nFor vowel sound only 2 sec audio is used from the vowel.wav file. So each of those have a shape of 1, 512
", "votes": null, "replies": [] } ] } ] } ], "raw_markdown_by_id": { "3027929": "jeevahimanshukaushik could you please clarify the model / encoder which were used to create the embeddings in the competition dataset? Thank you in advance.", "3027960": "samvelkoch it is generated using googles HeAR api. [Here](https://github.com/Google-Health/google-health/blob/master/health_acoustic_representations/README.md) is the github link for HeAR\n\nThe preprocessing for isolation of cough sounds was done using the code given in the overview. For vowel sounds, the mid interval from 2 secs to 4 secs was selected", "3028018": "Thank you 🙏 HeAR embeddings change the game bets. Very interesting to dive deeper.", "3028118": "Could you clarify a bit shapes of embeddings? Why cough embeddings are so different? Are their shape relate to cough files duration? ", "3028123": "jeevahimanshukaushik and dataset provenance if possible please. How did you collect the data?", "3028164": "Each sample of cough sound contains cough bouts ( a sudden burst of air for short interval in coughing). The shape of each of these samples are based out of number of cough bouts captured by the model. Interesting thing is we can consider each cough bout as a seperate data point as well to play with\n\nFor vowel sound only 2 sec audio is used from the vowel.wav file. So each of those have a shape of 1, 512", "3028167": "The data was collected using mobile phone via webapp. In most of the cases, recording was done on center. For a few of them recording was done remotely", "3058792": "Can you explain more detail about dataset source like the location (ex. Hospital name) or maybe the object of people, because I use this dataset for my college project?" }, "source": "meta" }