Talk:Speechify

COI edit request: voice AI platform and Simba models

Conflict-of-interest disclosure: I have a connection to Speechify (I work for the company). In line with WP:COI and WP:PSCOI, I am not editing the article directly and am instead requesting that uninvolved editors review and, if appropriate, incorporate the material below. I have kept the wording neutral and, wherever possible, cited independent reporting rather than the company's own press releases; please trim anything that still reads as promotional. (This request has been revised as I found stronger independent sources; an earlier outdated ranking figure was removed.)

Extended content

Summary of request

Between 2024 and 2026 Speechify broadened from a consumer text-to-speech reader into a wider voice-AI platform — adding voice cloning, a voice assistant, dictation, meeting transcription, and a proprietary voice-model family called Simba offered to developers through an API. The additions below are limited to events that received independent coverage (TechCrunch, 9to5Mac, OfficeChai). I am not requesting funding/valuation figures, university/partnership deals, celebrity-voice additions, award/"App of the Day" badges, or any "Voice Arena" ranking, because I could only find self-published or press-release sourcing for those.

Proposed additions

1. Recognition (suggested for the lead or a short "Recognition" line):

In 2025, Speechify received an Apple Design Award at Apple's Worldwide Developers Conference.

2. Product history (suggested as a short "History" or "Products" paragraph):

In February 2024, Speechify added Gmail integration and voice cloning to its iOS app. In November 2025 it added voice typing and a voice assistant to its Chrome extension. In early 2026 the company expanded its AI voice assistant, adding selectable celebrity voices and an integration with ChatGPT, along with meeting transcription and summarization. In March 2026 Speechify released a Windows application using on-device models for dictation and reading aloud, and reported more than 50 million users. In June 2026 it made voice typing free for all iPhone and Mac users.

3. Voice models (suggested as a sentence in the same paragraph):

Speechify develops a proprietary voice-model family called Simba, which it offers to third-party developers through a voice API. As of July 2026, Speechify's Simba 3.2 model held the top position (Elo 1233) on the independent Artificial Analysis Speech Arena text-to-speech leaderboard, ahead of Google's Gemini 3.1 Flash TTS; the leaderboard ranks models by blind listener preference and is updated continuously, so standings change over time.

Notes for reviewers

• Sourcing: every sentence above is cited to independent reporting (TechCrunch, 9to5Mac) except the benchmark ranking, which uses one independent write-up (OfficeChai) plus the leaderboard itself as a primary link so it can be verified directly. I have added the company's own press release on the ranking only as a clearly-labelled supplementary citation, not as the load-bearing source.
• On the ranking (item 3): the leaderboard is live and volatile, so I have written it as an attributed, dated statement. If reviewers prefer to omit a volatile ranking from an encyclopedia, I have no objection.
• Deliberately omitted for weak sourcing: funding/valuation (only conflicting aggregator figures exist; note the company's own materials variously say "50 million" and "60 million" users, so I have used the independently-reported "over 50 million"); any "Voice Arena" position (no independent coverage, and the live standing is a statistical tie rather than a clear rank); and the numerous company-announced items that appear only on Speechify's own news page (celebrity-voice additions, university access deals, individual investors, and "App of the Day"–type recognitions). If a reviewer finds independent coverage for any of these, they are welcome to add it.
• I have tried to follow WP:PROMO and would rather this be trimmed than overstated.

References

Thanks for taking a look. ~2026-38595-10 (talk) 14:21, 7 July 2026 (UTC)

 Not done: The {{request edit}} template is for requesting changes to semi-protected pages. For conflict of interest requests, please use {{Edit COI}} instead. I've updated the template for you. meamemg (talk) 14:22, 7 July 2026 (UTC)

Thanks Meamemg, appreciated — and sorry for the wrong template. One update since I posted: the ranking in item 3 now also has a company press release (added as a clearly-labelled supplementary citation only; the independent OfficeChai write-up and the leaderboard itself remain the primary support). I'm mindful this is a live, volatile leaderboard and a promotional claim, so I'm entirely happy for that sentence to be trimmed or dropped if you feel it doesn't belong in an encyclopedia. The recognition and product-history items (1 and 2) are the parts I'd most like considered, as they rest on independent reporting. ~2026-38595-10 (talk) 17:27, 7 July 2026 (UTC)

Update: I re-checked the live Artificial Analysis leaderboard and both the default and "selected voice" views now agree — Simba 3.2 is first at Elo 1233, with Gemini 3.1 Flash TTS second (1214). I've added the Elo figure to item 3 accordingly. (An earlier momentary discrepancy I mentioned appears to have been a stale cache.) ~2026-38595-10 (talk) 17:30, 7 July 2026 (UTC)

Reply 19-JUL-2026

  Edit request declined  

• The proposed text contains references from TechCrunch which act as press releases. (See WP:TECHCRUNCH.)

Regards,  Spintendo  23:35, 19 July 2026 (UTC)