Aqua Voice vs Google Voice Access

Aqua Voice is cloud-based dictation software that turns speech into text in everyday apps. It processes speech in real time, refining phrasing, grammar, and formatting, and supports 49 languages, custom vocabulary, and voice commands. Aqua claims 97.3% accuracy for its Avalon model on AISpeak. Starter is free, requires no credit card, and includes 1,000 words. Pro costs $8.00 per month billed annually and adds unlimited words, Custom Instructions, and an Expanded Custom Dictionary. Max costs $24.00 per month billed annually and adds Realtime Mode, the “Send it” voice command, and early feature access. Team costs $12.00 per month billed annually for 2–9 users; Business is for teams of 10 or more, with no price listed. Monthly and annual billing are available, and plans can be canceled anytime. Aqua lists Mac, Windows, and iOS availability; Android access is waitlist-only. Privacy Mode prevents transcripts being stored on Aqua’s servers. Business includes Zero Data Retention, while Team can enforce Privacy Mode.

Google Voice Access lets people control an Android device by speaking, including people who have difficulty manipulating a touchscreen. It can open apps, go back or home, show notifications and operate the screen currently in use. Users can display numbers or text labels beside screen items and speak the matching marker to select one. Voice Access also supports dictation and text editing, including replacing words or phrases. Google lists English, Spanish, German, Italian, French, Portuguese and Japanese; Portuguese and Japanese cover only the top 25 commands. Offline use requires downloading a speech-recognition language pack to the device. Speech is mainly processed on-device, but audio may be sent to Google servers depending on device settings and system language; Google says audio sent for this service is not saved. Audio and transcripts are not stored unless the user enables “Help improve Voice Access Commands.” The microphone listens while Voice Access is active. It does not identify a particular speaker, so anyone can control an awake, unlocked device while it is listening.