Google Nest Doorbells suffer mild identity crisis, start speaking with new accents (Updated)

The glitch emerged after Nest rolled out version 1.57 of the doorbell’s software, which introduced a new text‑to‑speech (TTS) library intended to make spoken alerts sound more natural. Instead of the standard American English voice that most U.S. customers expect, the system is intermittently selecting British, Australian, or even Indian English models when announcing “Someone’s at the door” or “Package delivered.” The issue is limited to the audible prompts; video, motion detection, and other core functions remain unaffected. Google has confirmed the problem, labeled it a “mild identity crisis,” and is pushing a hot‑fix to revert the TTS selection logic to the default accent while it investigates the root cause.

The incident underscores how tightly modern smart‑home hardware now relies on cloud‑driven AI components. Nest’s competitors—Amazon’s Ring Doorbell and Apple’s HomePod‑linked doorbell concepts—also use AI‑generated speech, but they have not reported similar mis‑routing of voice profiles. Google’s push to unify its Assistant voice across devices means any misstep can ripple through multiple product lines, eroding the polished, homogeneous brand experience that has been a selling point for Nest. Moreover, the glitch arrives at a time when consumers are increasingly sensitive to the authenticity of AI interactions; a doorbell that sounds “foreign” to a user can feel jarring and may even raise concerns about privacy or misconfiguration.

What follows will be a rapid rollout of a corrective firmware patch, likely within the next 48 hours, and a more thorough audit of the TTS deployment pipeline. Developers should watch for any spill‑over of the accent‑switching bug into other Nest products that share the same voice engine, such as Nest Cam or Nest Hub. Additionally, the episode may prompt Google to add a user‑selectable voice setting for doorbells, giving owners control over regional accents and reducing reliance on a single default model. Monitoring user sentiment on forums and social media will be key to gauge whether the temporary annoyance translates into churn or brand‑trust erosion.