Wikipedia:Peer review/15.ai/archive1

15.ai

Article (edit | visual edit | history) · Article talk (edit | history) · Watch • Watch peer review

I've listed this article for peer review because it has a potential to be a decent article, but it has had a difficult history. Namely, it was affected by socking earlier on, causing it to improperly become a GA; then it underwent a GAR and was even deleted. In the recent discussions, primarily the last AfD, there was much discussion around the strength of sourcing. I am interested in an outside reviewer's position primarily on whether certain statements supported by some of the weaker sources should be removed. The second issue is the appropriateness of the 'Background' section, seeing how most of the statements there come from sources that are not about the subject.

Thanks, —Alalch E. 23:14, 28 December 2024 (UTC)[reply]

I'm on my phone, so I'm not in a position to be evaluating source quality. However, I do want to affirm that the background section is too detailed. This is not an article about 15.ai, not an article about speech generation generally. I think that section could be literally one paragraph that describes the shift from pre-2016 to DeepMind. At the end of it, put an "Also Tacotron was big at the time." If Microsoft FastSpeech was not a significant influence on 15, it does but belong here. It's not clear to me what the final paragraph even currently adds. After that one paragraph, you should be good to launch into the actual subject of the article: 15.ai. lethargilistic (talk) 04:00, 9 January 2025 (UTC)[reply]

@Lethargilistic: Thank you. Would you kindly take a look, probably when you're on a computer, at the issues recently (after this peer review request was posted) identified by User:Emm90 in Special:Diff/1268071188, and which were either addressed or dismissed by User:GregariousMadness as seen in the following diff: Special:Diff/1268071188/1268086148. I am noting that many tags were removed without being resolved in the way intended by that tag, so for example, www.equestriacn.com was tagged as an unreliable source, and the tag was simply removed. Provided that you've reviewed the tags placed by Emm90, and how they were addressed and not addressed, would you say that as a whole these issues have been addressed appropriately or not quite? —Alalch E. 13:31, 9 January 2025 (UTC)[reply]

What I could do is move most of that background section into the article Deep learning speech synthesis article and summarize the key points in the 15.ai background section. Would that be okay? GregariousMadness (talk to me!) 14:31, 9 January 2025 (UTC)[reply]

Yes please. —Alalch E. 14:36, 9 January 2025 (UTC)[reply]

Done! I moved it over to deep learning speech synthesis. On that note, I reached out to the people who wrote the ElevenLabs blog post, and I found out that the CEO and founder of ElevenLabs himself (Mati Staniszewski) was the one who wrote the blog post (you can verify by playing the audio file, attached in the blog — it says his name). I'll add a short sentence about this as well. GregariousMadness (talk to me!) 16:19, 10 January 2025 (UTC)[reply]