WhatsApp is developing a dictation tool for Android that aims to change how messages are sent without using voice notes. The idea is simple: the user speaks into the phone, the app transforms those words into text, and the result is sent as a regular text message. The recipient wouldn't receive any audio or see a label indicating that the message was written using voice.
The new feature has been spotted in the Android beta version 2.26.37.8, but its presence in the code doesn't mean it's available for testing. It remains under development, is not yet available to beta users, and WhatsApp hasn't announced a release date . It's also unknown whether it will be released the same way on iPhone or if it will change before its official launch.
A new way of writing by speaking
According to the implementation found, the app will place a new button in the chat's text input bar. The first time it's used, it will display a brief explanation and ask the user to choose a language. Afterward, it will enter listening mode and display the message "Listening..." while converting the voice into words. When the dictation is complete, the user can review the text before sending it.
If the conversation contains errors or the user decides not to send the message, there will be an option to cancel it. The feature will work in individual conversations, groups, and channels, although the design and these destinations may still change. The button's location is not final , and you don't need to activate anything in Settings to find it: those using the beta version may not see it because the code is still being developed.
It's important not to confuse this with the existing audio transcription feature. WhatsApp allows you to convert a received voice note to text, but in that case, the audio remains part of the chat, and the transcription appears below. The new feature works before sending the message and in reverse: the voice serves as the input method but disappears afterward . This prevents the recipient from having to listen to a recording, and since it remains as text, an address, name, or any other information can be searched for later within the conversation.
It also differs from the microphone built into Gboard and other keyboards. The result may be similar, but WhatsApp wants to integrate it into its own toolbar and leverage the same infrastructure it already uses for transcriptions. The process would remain within the application and offer a consistent experience even if the installed keyboard changes.
Privacy, language, and outstanding questions
The conversion will be done locally. WhatsApp will prompt you to download the chosen language pack once, and from then on, it will be able to process the dictation offline. You'll only need to reconnect to send the message to the chat. The temporary recording will not leave the device , according to available information.
The resulting text will travel like any other WhatsApp message and will be protected by end-to-end encryption in conversations that use it. This is the same approach the company applies to official voice message transcripts on Android and iPhone: a package is downloaded and the content is processed on the device itself.
Recognition is not infallible. Noise, rapid speech, proper names, or mixing languages ​​can introduce errors. Herein lies a significant advantage: the sender can correct them before sending. This preliminary review reduces the risk of sending a faulty text.
WhatsApp tests features that sometimes change their appearance, location, or never reach the stable version. It's also unknown which languages ​​will be available on day one or if the rollout will extend to iPhones simultaneously. For now, there's no need to activate anything or look for the option in Settings . When the company decides to test it publicly, it will typically enable it first for a small group of accounts.
The proposal fills a different gap than apps and bots that transcribe WhatsApp audio to text. It would allow users to speak when their hands are busy or quickly write a long message without burdening the recipient with the time and obligation of listening to a recording. If it is eventually released, the user will be able to dictate, review, and send a plain text message, with local processing and without notifying the recipient. For now, we have to wait , because being in beta doesn't mean immediate availability.