What’s new in VelvetSpeak v1.3.6

VelvetSpeak v1.3.6 brings Solo Help: a way to get useful wording even before the other person has spoken. It also builds on a busy run of updates since our last news post about v1.2.1, with better conversation context, more ways to refine a reply, clearer answer previews and steadier listening.

Solo Help: find your words before the conversation gets going

Previously, swiping down for help when only your own voice had been heard could leave you with “Not enough said yet” or “Only ‘You’ so far”.

With Solo Help, say a full sentence about what you want to express, then swipe down. Velvet gives you a Try saying card with a stronger way to put your latest words. This can help you practise an introduction, make a request clearer or find a better way to start a difficult conversation.

Different answer gives you fresh wording, and Guide me is available if you want guidance. Once the other person has spoken, ordinary conversation Help takes over.

Solo Help arrived in v1.3.5. v1.3.6 fixes an important edge case: the glasses sometimes leave a lone speaker’s words unattributed. Those unassigned lines no longer prevent Solo Help when nothing on record is labelled as the other person. If another person has been identified, Velvet keeps the normal conversation-help path.

Before anyone speaks, the glasses now say “Say a sentence · ↓ get help”. If your speech is too short or still being transcribed, they tell you to say a full sentence or try again once listening catches up. A successful Solo Help request uses your suggestion allowance, just like other Help requests.

More useful help throughout a conversation

Keep the conversation when you change lenses — v1.3.0. Switching Lens or Scene during a session now carries the conversation and its running summary forward. Your next request can still use what was just said, while the new lens loads its own preparation. You can move from an everyday chat to a work situation without starting from an empty conversation.

More useful preparation and recaps — v1.3.0–1.3.1. The phone can list and activate Plans saved on the dashboard, and About me gives you a place to supply personal context. Updated recaps can show what the other person shared and a copyable follow-up suited to the lens. Saving, activating and using preparation are made clearer, including messages when an action cannot be completed.

Refine a suggested line — v1.3.1. Supported “What can I say?” results offer New approach and lens-specific options such as Warmer, Funnier, More confident or Shorter. You can ask for a different move or adjust the wording without changing your saved Style. These options depend on the lens and the type of result; each refinement is another suggestion request.

See an answer before opening it — v1.3.2. With the transcript off, a ready answer can show a short preview of its opening sentences on the listening screen. The preview tells you what is waiting and how to open it. It uses whole sentences rather than cutting a line off, and an answer that arrives while you are reading a card or using a menu goes into Recent instead of interrupting you.

Better-grounded suggestions — v1.3.3

Velvet now keeps more of the current conversation in view, helping it bring back relevant details from earlier.

We also tightened the rules around personal facts. Suggested lines are directed to use your About me, your Plan or things you actually said, rather than inventing details about your family, interests or experiences. When the information is missing, Velvet should ask back instead.

Placeholder wording such as “[band name]” and lines cut off mid-word are filtered out. Saved facts about the person you are talking to are selected for relevance to the current conversation.

These changes target common sources of awkward suggestions. They are not a guarantee that every answer will be right, so specific examples of what helped or missed the mark remain useful.

Steadier listening when the connection drops — v1.3.4

Reconnects now start in about a second instead of waiting around 15 seconds. Velvet keeps listening during slow uploads and avoids sending background noise while catching up, reducing the speech that can be lost during a connection problem.

The “Reconnected · missed ~N s” notice reports measured gaps, and the planned hourly reconnect happens quietly when it succeeds. A connection drop can still leave a gap; the notice helps make that visible.

Try it and tell us how it went

When v1.3.6 is available to your account in Even Hub, try saying a sentence while you are the only person speaking, then swipe down. During a conversation, try changing lenses or using New approach on a suggested line.

Some features depend on server availability and rollout settings, so installing an app version does not enable every feature at once.

After a session, use Send feedback on your phone to tell us about one suggestion that helped—or one that should have. You can send feedback without attaching conversation text; optional excerpts are previewed and shared only when you choose them.