Apologies for the naive question, but just wanted to make sure.
\nThe human voices on the test sets are not related to the labels correct (unlike the training data where in several cases they do describe the species heard) ??
", "messages": [ { "id": "3162086", "postDate": "03/28/2025 19:27:28", "content": "Apologies for the naive question, but just wanted to make sure.
\nThe human voices on the test sets are not related to the labels correct (unlike the training data where in several cases they do describe the species heard) ??
", "rawMarkdown": "Apologies for the naive question, but just wanted to make sure. \n\nThe human voices on the test sets are ***not*** related to the labels correct (unlike the training data where in several cases they **do** describe the species heard) ??", "votes": null }, { "id": "3162139", "postDate": "03/28/2025 20:43:29", "content": "Not sure we can say since we don't every get to see the test data. The test soundscapes are stated to be similar to the train soundscapes - when you listen to that speech content does it describe the species? I have only listened to train_audio stuff so far myself and don't speak spanish.
", "rawMarkdown": "Not sure we can say since we don't every get to see the test data. The test soundscapes are stated to be similar to the train soundscapes - when you listen to that speech content does it describe the species? I have only listened to train_audio stuff so far myself and don't speak spanish.", "votes": null }, { "id": "3183148", "postDate": "04/20/2025 12:43:36", "content": "@stefankahl can you please confirm this is not the case??
\nthank you
", "rawMarkdown": "stefankahl can you please confirm this is not the case??\n\nthank you", "votes": null }, { "id": "3183289", "postDate": "04/20/2025 17:22:40", "content": "There may be some incidental human voice in the test set. This will generally not be similar to the descriptive notes in some training examples.
\n(Training data is usually what we call focal recordings - someone pointing a microphone at a particular animal - and sometimes recordists include some voice notes about what they're hearing and the surrounding context. The test data, on the other hand, is passive acoustic data, meaning that the sounds captured are entirely ambient. People may wander by, or even say something when setting up the microphone, but this is rarely descriptive of the immediate sounds.)
", "rawMarkdown": "There may be some incidental human voice in the test set. This will generally not be similar to the descriptive notes in some training examples.\n\n(Training data is usually what we call focal recordings - someone pointing a microphone at a particular animal - and sometimes recordists include some voice notes about what they're hearing and the surrounding context. The test data, on the other hand, is passive acoustic data, meaning that the sounds captured are entirely ambient. People may wander by, or even say something when setting up the microphone, but this is rarely descriptive of the immediate sounds.)", "votes": null } ], "comments": [ { "id": 3162139, "author_name": "pcjimmmy", "author_url": "", "post_date": "03/28/2025 20:43:29", "content": "Not sure we can say since we don't every get to see the test data. The test soundscapes are stated to be similar to the train soundscapes - when you listen to that speech content does it describe the species? I have only listened to train_audio stuff so far myself and don't speak spanish.
", "votes": null, "replies": [] }, { "id": 3183148, "author_name": "ikarosilva", "author_url": "", "post_date": "04/20/2025 12:43:36", "content": "@stefankahl can you please confirm this is not the case??
\nthank you
", "votes": null, "replies": [ { "id": 3183289, "author_name": "tomdenton", "author_url": "", "post_date": "04/20/2025 17:22:40", "content": "There may be some incidental human voice in the test set. This will generally not be similar to the descriptive notes in some training examples.
\n(Training data is usually what we call focal recordings - someone pointing a microphone at a particular animal - and sometimes recordists include some voice notes about what they're hearing and the surrounding context. The test data, on the other hand, is passive acoustic data, meaning that the sounds captured are entirely ambient. People may wander by, or even say something when setting up the microphone, but this is rarely descriptive of the immediate sounds.)
", "votes": null, "replies": [] } ] } ], "raw_markdown_by_id": { "3162086": "Apologies for the naive question, but just wanted to make sure. \n\nThe human voices on the test sets are ***not*** related to the labels correct (unlike the training data where in several cases they **do** describe the species heard) ??", "3162139": "Not sure we can say since we don't every get to see the test data. The test soundscapes are stated to be similar to the train soundscapes - when you listen to that speech content does it describe the species? I have only listened to train_audio stuff so far myself and don't speak spanish.", "3183148": "stefankahl can you please confirm this is not the case??\n\nthank you", "3183289": "There may be some incidental human voice in the test set. This will generally not be similar to the descriptive notes in some training examples.\n\n(Training data is usually what we call focal recordings - someone pointing a microphone at a particular animal - and sometimes recordists include some voice notes about what they're hearing and the surrounding context. The test data, on the other hand, is passive acoustic data, meaning that the sounds captured are entirely ambient. People may wander by, or even say something when setting up the microphone, but this is rarely descriptive of the immediate sounds.)" }, "source": "meta" }