Google is back with another AI service — this time, an offline dictation program using its “Gemma” architecture. But instead of including it in the Gemini app or as a Gemini function, the company has decided to roll it out in a dedicated iPhone app, with too much catchy name”Google AI Edge Eloquent”
I decided to give the app a shot on release day, though the privacy policy gave me pause. Google says your location, contacts, identifiers, device diagnostics, contact information, user content, usage data and “other” data can be linked to you, while purchases and other diagnostics can be collected but not linked to you. That’s a lot of data, especially for an app that advertises that “audio, private conversations, and personal data never leave your device,” and I’m not sure I’d be eager to download the app otherwise. But, as the saying goes, if the service is free, you is the product. I’ve reached out to Google for clarification here and will update this story if I hear back.
How to try Google’s new AI transcription app
Once you’ve downloaded the app, setup is easy—you record a sample example sentence the app asks you to say, then choose: “Mode on device,” which is completely offline, and stores your conversations online on your device; or “advanced text polishing,” which keeps Audio on your device, but uses Gemini to “polish” your text, which requires you to send the data to the cloud (and presumably where the aforementioned privacy policy data is going). You won’t need to have Gemini running to do basic editing of your transcript—though by design, the app removes “filler” words like “um.” Note that the app seems to open in “advanced text polishing” mode by default—at least, that’s how it works on my end. But a simple tap of the toggle in the top right corner of the main screen switches you into “On Device Mode”.
I had some trouble getting the app up and running: every time I tried to test it, it claimed I wasn’t speaking at all. But after pairing and unpairing the AirPods with my iPhone, the app seemed to work. To test the app, I played the intro of The Audio University YouTube videoWhich is completely dialogue based. Once the app was up and running, it immediately began transcribing the video with near-perfect accuracy—at least by the end. I would watch the app enter the wrong words, then back out and replace the subsequent words as given context. Once the recording was finished, the transcript was almost identical to the video transcript, save for a few quirks: he mistakenly thought “if this is our first time meeting” as “this is our first time meeting,” and recorded a sentence twice. But other than that, this is a pretty useful transcript of the beginning of the video.
What do you think so far?
From here, you have many options—especially if you invite Gemini to help. If you want to correct any text that the AI ”polished” misspelled, you can tap the pencil icon on the transcript to manually edit it. Above this, you can see “usage statistics” including the number of words spoken, words spoken per minute, and the number of edits made by AI. If you switch to Gemini, you’ll have access to additional AI editing tools, including “Key Points,” “Formal,” “Short,” and “Long.” When you are satisfied with the transcription, you can tap the Copy button to paste the text elsewhere on your clipboard. In the “History” tab, you can view your previous transcriptions and return to them to edit them (manually or with AI). In the “Dictionaries” tab, you can add obscure words that you use frequently but the AI might not turn on, further improving the accuracy of your recordings.
In my brief testing, the app worked well, and I appreciate the option to use it only on the device. I’d definitely consider using it over iOS’s built-in transcriptions if it seemed faster or more accurate, especially since there are some more robust features here – assuming that’s actually on the device. does It means keeping my data out of Google’s hands.
