![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F3623956%2F0bc412499aef13d2d21470ed85dca790%2FCaptura%20desde%202025-10-14%2019-00-27.png?generation=1760461256839904&alt=media\")
Is this competition to be launched next year as well?
\nBy the way, as the image and the problem-related video shows, the subject sees the image of the text to be read out loud during intending to say those words. How confident can we be, that the readings that we have right now are correlated much stronger with the intention of saying those words than it is correlated with the text being seen on the display device? Do SOTA solutions work as well without - the person being recorded - seeing words on a screen?
\n
\n\nIs this competition to be launched next year as well?
\n
We are not certain if this competition would be run again next year, as it would require a new dataset. Stay tuned.
\n\n\nBy the way, as the image and the problem-related video shows, the subject sees the image of the text to be read out loud during intending to say those words. How confident can we be, that the readings that we have right now are not correlated much stronger with the intention of saying those words than it is with the text being seen on the display device?
\n
The participant also uses the device in an unprompted mode for communication. In this mode, no \"prompt\" text is shown on the screen, he is simply attempting to say the words that he wants to communicate, and the decoding still works quite well. So we are quite confident that the neural modulations are related to the intention to speak and not with reading or listening to prompted text. See:
\n\n\nIs there any data with people speaking? shouting/normal/whispering etc. ? a microphone, recording voice could be GT
\n
We have a seperate study about how volume is encoded in neural signals. The preprint is available here: https://www.biorxiv.org/content/10.1101/2025.05.30.657105v1.abstract
", "rawMarkdown": "> Is this competition to be launched next year as well?\n\nWe are not certain if this competition would be run again next year, as it would require a new dataset. Stay tuned.\n\n> By the way, as the image and the problem-related video shows, the subject sees the image of the text to be read out loud during intending to say those words. How confident can we be, that the readings that we have right now are not correlated much stronger with the intention of saying those words than it is with the text being seen on the display device?\n\nThe participant also uses the device in an unprompted mode for communication. In this mode, no \"prompt\" text is shown on the screen, he is simply attempting to say the words that he wants to communicate, and the decoding still works quite well. So we are quite confident that the neural modulations are related to the intention to speak and not with reading or listening to prompted text. See:\n- https://www.nejm.org/doi/full/10.1056/NEJMoa2314132\n- https://www.biorxiv.org/content/10.1101/2025.06.26.661591v1.abstract\n\n> Is there any data with people speaking? shouting/normal/whispering etc. ? a microphone, recording voice could be GT\n\nWe have a seperate study about how volume is encoded in neural signals. The preprint is available here: https://www.biorxiv.org/content/10.1101/2025.05.30.657105v1.abstract", "replies": [ { "id": 3303184, "postDate": "2025-10-17T10:12:41.707Z", "content": "Thank you for the detailed response!
\nIf the device works well without seeing a prompt, that's reassuring!
\nAbout the additional data, I am curious if you recorded data where people with no ALS use the device while speaking - recording EEG while a microphone picks up their voice. I am curious because there might be some valuable transfer learning from predicting spoken words (or maybe even EEG to Speech directly) to intended spoken words, helping the decoding of intended ones.
\nAlso, a successful direct EEG to Speech solution might make participants able to learn how to control tone, loudness and such intricacies of speech. Additionally, they could set a speech embedding for how they would like their decoded voice to sound like.
I have found this competition through ChatGPT, looking for EEG datasets. But I have realised this competition is not EEG-based, sorry for the misunderstanding!
", "rawMarkdown": "I have found this competition through ChatGPT, looking for EEG datasets. But I have realised this competition is not EEG-based, sorry for the misunderstanding!" } ] }, { "id": 3513838, "postDate": "2026-08-17T18:52:40.127Z", "content": "Hi there! Is there any update on wether the competition will be held again in the future?
", "rawMarkdown": "Hi there! Is there any update on wether the competition will be held again in the future?" } ] }, { "id": 3301977, "postDate": "2025-10-14T17:07:47.813Z", "content": "Is this competition to be launched next year as well?
\nBy the way, as the image and the problem-related video shows, the subject sees the image of the text to be read out loud during intending to say those words. How confident can we be, that the readings that we have right now are correlated much stronger with the intention of saying those words than it is correlated with the text being seen on the display device? Do SOTA solutions work as well without - the person being recorded - seeing words on a screen?
\n
\n\nIs this competition to be launched next year as well?
\n
We are not certain if this competition would be run again next year, as it would require a new dataset. Stay tuned.
\n\n\nBy the way, as the image and the problem-related video shows, the subject sees the image of the text to be read out loud during intending to say those words. How confident can we be, that the readings that we have right now are not correlated much stronger with the intention of saying those words than it is with the text being seen on the display device?
\n
The participant also uses the device in an unprompted mode for communication. In this mode, no \"prompt\" text is shown on the screen, he is simply attempting to say the words that he wants to communicate, and the decoding still works quite well. So we are quite confident that the neural modulations are related to the intention to speak and not with reading or listening to prompted text. See:
\n\n\nIs there any data with people speaking? shouting/normal/whispering etc. ? a microphone, recording voice could be GT
\n
We have a seperate study about how volume is encoded in neural signals. The preprint is available here: https://www.biorxiv.org/content/10.1101/2025.05.30.657105v1.abstract
", "votes": 0, "replies": [ { "id": 3303184, "author_name": "nofreewill42", "author_url": "", "post_date": "2025-10-17T10:12:41.707000", "content": "Thank you for the detailed response!
\nIf the device works well without seeing a prompt, that's reassuring!
\nAbout the additional data, I am curious if you recorded data where people with no ALS use the device while speaking - recording EEG while a microphone picks up their voice. I am curious because there might be some valuable transfer learning from predicting spoken words (or maybe even EEG to Speech directly) to intended spoken words, helping the decoding of intended ones.
\nAlso, a successful direct EEG to Speech solution might make participants able to learn how to control tone, loudness and such intricacies of speech. Additionally, they could set a speech embedding for how they would like their decoded voice to sound like.
I have found this competition through ChatGPT, looking for EEG datasets. But I have realised this competition is not EEG-based, sorry for the misunderstanding!
", "votes": 0, "replies": [] } ] }, { "id": 3513838, "author_name": "nofreewill42", "author_url": "", "post_date": "2026-08-17T18:52:40.127000", "content": "Hi there! Is there any update on wether the competition will be held again in the future?
", "votes": 0, "replies": [] } ] }, { "id": 3301978, "author_name": "", "author_url": "", "post_date": "2025-10-14T17:14:56.613000", "content": "", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "3301985": "> Is this competition to be launched next year as well?\n\nWe are not certain if this competition would be run again next year, as it would require a new dataset. Stay tuned.\n\n> By the way, as the image and the problem-related video shows, the subject sees the image of the text to be read out loud during intending to say those words. How confident can we be, that the readings that we have right now are not correlated much stronger with the intention of saying those words than it is with the text being seen on the display device?\n\nThe participant also uses the device in an unprompted mode for communication. In this mode, no \"prompt\" text is shown on the screen, he is simply attempting to say the words that he wants to communicate, and the decoding still works quite well. So we are quite confident that the neural modulations are related to the intention to speak and not with reading or listening to prompted text. See:\n- https://www.nejm.org/doi/full/10.1056/NEJMoa2314132\n- https://www.biorxiv.org/content/10.1101/2025.06.26.661591v1.abstract\n\n> Is there any data with people speaking? shouting/normal/whispering etc. ? a microphone, recording voice could be GT\n\nWe have a seperate study about how volume is encoded in neural signals. The preprint is available here: https://www.biorxiv.org/content/10.1101/2025.05.30.657105v1.abstract", "3301977": "Is this competition to be launched next year as well?\n\nBy the way, as the image and the problem-related video shows, the subject sees the image of the text to be read out loud during intending to say those words. How confident can we be, that the readings that we have right now are correlated much stronger with the intention of saying those words than it is correlated with the text being seen on the display device? Do SOTA solutions work as well without - the person being recorded - seeing words on a screen?\n\n\n", "3301978": "" } }