/Tech2h ago

Gemini Flash Upgrades Mic with 70+ Language Support on Mobile

52343252218.3K

#1238

Original post

Josh Woodward@joshwoodward#1238inTech

Our mic icon just got way better on Android and iOS! This is HUGE for non-English speakers.

+ now 70+ languages + mix languages freely + don't have to change language settings + still doesn't interrupt you :)

5:03 PM · Jun 15, 2026 · 16.4K Views

Sentiment

Positive users hail Gemini Flash's mic upgrade for adding 70+ language support and better accessibility, while negative users complain about persistent voice bugs and lost trust in reliability.

Pos

71.4%

Neg

28.6%

22 comments with sentiment.

Cluster Engagement

Posts from X

Most Activity

VIEWS2KLIKES38REPLIES4

Josh Woodward@joshwoodward

And coming to the Web in another week or so!

Josh Woodward@joshwoodward

Our mic icon just got way better on Android and iOS! This is HUGE for non-English speakers.

+ now 70+ languages + mix languages freely + don't have to change language settings + still doesn't interrupt you :)

2h2K382

BOOKMARKS3

Art Seabra@ifthis

Josh, this is exactly the seam I’ve been trying to name. I have been documenting this frustration inside my workflow for a bit now. I'd be happy to share it all with you and team. =]

My frustration has not been “voice is bad.” It’s that voice-only workflows often become the substrate instead of compiling from one.

Once the voice layer is the thing carrying memory, notation, edits, citations, and intent, copy/paste downstream gets below my risk tolerance fast.

The solution I’ve been building is closer to voice-mostly than voice-first: source-of-truth text/metadata stays preserved, then audio is compiled from it with read-as notation, structural cueing, captions/transcripts, and bounded risk surfaces.

So yes — I’ll try it again on a low-risk wave and send you the notes.

2h273

Sreeraman Thiagarajan@sreeraman

@joshwoodward Take Josh out of NBU, but not NBU out of Josh 🥳

1h1111

Josh Woodward@joshwoodward

@sreeraman More coming along these lines, too :)

1h951

EPOXY EXPERT DESIGN@design65517

@joshwoodward Does Mr. Josh plan to include in the NotebookLM audio overview the ability to answer and ask questions in any language?

2h281

Art Seabra@ifthis

@joshwoodward not sure i can go back to trusting it Josh. i've lost soooo many vocal thoughts because of all, not just gemini's. Voice Memos as the default action button on my Watch Ultra ⊹ transcribe ⊹ copy/paste downstream has proven to fail below risk tolerance.

2h17

Diablo Loco@SambaD23222

@joshwoodward I am working on data collection for bambara and nko. http://Nkodon.com

32m7

WK@zoom_will

@joshwoodward Can you teast when Gemini 3.5 Pro?

2h1101

Aziz@azizlanou

@joshwoodward Just tried it! final it actually works! can't wait to have this on Web.

1h2

miguel@miguevf1908

@joshwoodward At last ! We need it , and we need that changes ! Thanks

2h731

Federico Lix Klett@FedericoLKG

@joshwoodward Al fin!!!!! Era una de las funciones que mas pedía.

Vamos! Ya lo voy a testear.

2h182

Bourke Floyd, IV@BourkeFloyd

@joshwoodward @GeminiApp I would give it a cool animated icon with all that effort 😎❤️

1h561

𝙇𝙐𝙄𝙎 𝕏⃠@LuisSnuis

@joshwoodward I wish the overlay/"hey google" one was as smart, that one appears to just be some old basic speech to text one? I may be wrong, but it is very bad at recognition, especially if you have more than one language added

2h371

Mohammed Hossam@MohamedHz72007

@joshwoodward Gemini’s voice is worse than ChatGPT’s and Claude’s—so poor I may dictate a prompts to ChatGPT to past to gemini. After minutes of speaking, only one English word transcribed to Arabic, likely due to initial language detection, frustrated me and broke trust until now.

1h321

EPOXY EXPERT DESIGN@design65517

@joshwoodward This would be incredible for the NotebookLM, both for the audio generated in the NotebookLM and for the live chat function to communicate with all the Notebook's sources.

2h301