أجزاء من هذه الصفحة غير متاحة بعد بهذه اللغة، وتُعرض بالإنجليزية.
What is being built now
My Words is in use today as a written conversation. These are the next pieces of work, and they are included for everyone on a current plan when they land.
Speaking and listening, with the tools you already have
My Words has no speech of its own yet. What it does have is ordinary text fields, and those work with whatever your device already uses to put words into one.
- Dictation puts speech in. The built-in dictation on a phone, tablet or computer types into a My Words turn the same way it types into an email. Nobody has to learn anything new, and it is the person's own device doing the listening.
- A screen reader reads turns out. Each turn is ordinary text with its language declared on it, so a screen reader announces a translated turn in the voice for the language it is actually in.
- It works on both sides of the conversation. The person joining on their own phone gets the same, which matters where somebody reads their own language slowly.
Speech to text and text to speech built into the product are still worth doing, because that is where accents, background noise and the scoring of a spoken turn get handled properly. This is what is available in the meantime.
Not independently tested yet
The above follows from how the product is built rather than from a test we have run and published. No formal audit against WCAG 2.2 AA has been carried out, and we have not tested every combination of dictation tool, screen reader and browser.
If you rely on a particular one, tell us which and we will check it against yours. What has and has not been checked is set out on the accessibility page.
What is being built now
My Words is improved continuously, and every improvement reaches you automatically. It runs in the browser, so there is no version to install, nothing to migrate and no upgrade to buy: the next time you open a conversation, you are on the current build. These are the next pieces of work, and they arrive on every plan as they land.
Speech in
Speech to text, so a person can talk instead of type and the turn is transcribed in their own language before it is translated. The check, the score and the clarifying question all work the same way on a spoken turn.
Speech out
Text to speech, so a translated turn can be heard as well as read. This matters most where someone reads their own language slowly, or not at all.
Document translation
Translating the letter, form or notice that the conversation is actually about, with the same context and the same per-item scoring as a conversational turn.
More languages
Arabic, Urdu, Bengali, Somali, Farsi and Mandarin, with the right-to-left and script-specific work that goes with them.