Edgars Adamovics

DevRel @Speechmatics. I build voice agents and speech demos for real-world audio. Code, AI, benchmarks, and the occasional talking game. Start building below.

[Edgars Adamovics @BinaryCoder]

Building voice AI? I work in DevRel at @Speechmatics.

I share runnable demos, integration notes and tests with real-world audio from voice agents to talking games.

Start here: https://t.co/Dlr0NWh8Pi

What are you building, and where does speech fail?

[Edgars Adamovics @BinaryCoder]

@NoMansSky Will it run on iphone duo in full screen @NoMansSky ? Please say yes!!!

[Kai Lentit (e/xcel) @KaiLentit]

Hyperoptimized Engineer.

[Jason Lengstorf @jlengstorf]

CASTING CALL: I had a team drop out, so I need 2 devs on short notice to join an episode of Web Dev Challenge in Portland on Oct 21–23

TimescaleDB and sensor data challenge. if that sounds like fun, grab a friend and slide into my DMs ASAP! 💜

[Edgars Adamovics @BinaryCoder]

@Speechmatics

[Edgars Adamovics @BinaryCoder]

Building voice AI? I work in DevRel at @Speechmatics.

I share runnable demos, integration notes and tests with real-world audio from voice agents to talking games.

Start here: https://t.co/Dlr0NWh8Pi

What are you building, and where does speech fail?

[Edgars Adamovics @BinaryCoder]

@Speechmatics

[Edgars Adamovics @BinaryCoder]

@Speechmatics

[Edgars Adamovics @BinaryCoder]

Building Voice AI for healthcare? Here is a challenge worth thinking about.

What happens when a doctor mentions a complex medication, a patient switches languages mid-sentence, and multiple people are speaking?

Your STT needs to keep up.

I have been looking at our new Oak model at @Speechmatics , and the benchmarks are interesting:

91% medical term recall, ranking #1 across all 11 languages tested against 8 other providers.

If you are curious to experiment, we have released an Oak Consultation Studio example on GitHub:

https://t.co/iTuCnJ3KIi

Would love to hear from developers building medical scribes or healthcare agents. What's been your biggest challenge with speech recognition?

#VoiceAI #HealthcareAI #DevRel

[Speechmatics @Speechmatics]

Speech providers mistranscribe around 1 in 4 medical terms. ❌

That's why we built Oak.

Introducing our new multilingual medical speech-to-text model. 🌳

[Edgars Adamovics @BinaryCoder]

12k for 5090 6k for standard...

Ok this is slightly out of control now...

https://t.co/3ZFTzyoK1y

[Amanda @hey_amandam]

We are officially one month out from VapiCon. If you are building with voice this is the room you wanna be in to connect with leaders pushing what's possible.

Who will I see there?

https://t.co/OObwCJGRe6

[Edgars Adamovics @BinaryCoder]

@AkritiKeswani Real devs do it in bathroom just saying...

[Edgars Adamovics @BinaryCoder]

@hey_amandam @chantastic @Vapi_AI I need to get hold of one too ^^

[Edgars Adamovics @BinaryCoder]

I think the challenges are becoming much more about systems than models now.

Things like robust evals for real conversations instead of lab audio, knowing when to interrupt, wait or clarify, handling state across long messy calls, observability when something goes wrong, safe tool use and graceful failure paths.

There is also the challenge of keeping quality high across accents, domains and telephony while still making the economics work at scale.

We have made huge progress on whether a voice agent can work. The harder question now is whether it can work reliably 100,000 times a day in production.

[Edgars Adamovics @BinaryCoder]

After a week with @livekit @fin_ai , @aquavoice ,@livekit, @cartesia , @AgoraIO, @Vapi_AI and many other builders across SF, one thing is clear:

Voice AI has moved past can it talk?

The real question is whether it understands well enough to act.

Try @Speechmatics. Stress it. Break it. Show us what you build.

Here is my summary below 👇

#SFTechWeek #VoiceAI

[Edgars Adamovics @BinaryCoder]

@bitforth @adelwu_ Because some of us has families, different priorities and we don't have time to dedicated to grow business built on feeling that it might make it.

[Edgars Adamovics @BinaryCoder]

@adelwu_ You just not looking in right places...

[Edgars Adamovics @BinaryCoder]

@BenDenny @thorwebdev Those were bad companies who did not understand what development does.

[Edgars Adamovics @BinaryCoder]

@hey_amandam @livekit @fin_ai @aquavoice @cartesia @AgoraIO @Vapi_AI @Speechmatics Panel was great! I see you at Vapicon!!!

[Edgars Adamovics @BinaryCoder]

@Speechmatics vs the Golden Gate Bridge vs F-15s

One laptop.
Built-in mic.
No cleanup.
Wind, traffic, crowds, jets overhead.
I ran live transcription anyway.

Because STT doesnt just need to work in a quiet room.

It needs to get names, instructions and numbers right when the real world is doing everything it can to break it.

That’s the test.
#Speechmatics #VoiceAI #SpeechToText #SFTechWeek

[Edgars Adamovics @BinaryCoder]

@realharismalik @aquavoice @livekit @Speechmatics True.. And that is a part of problem even most acurate transcript cannot account for human stupidity when under influence of alcohol.

You think you saying right things but in reality you said completely something different and transcript just did its job.

[Edgars Adamovics @BinaryCoder]

The more powerful voice agents become, the less room there is for close enough.

Great session tonight at #SFTechWeek with @aquavoice , @livekit and @Speechmatics .

The demos were fun, but the bigger point was serious:

If a voice agent can change your calendar, update systems, or cancel a flight, one bad transcription can have real consequences.

That is why accuracy, context, turn-taking and diarization matter just as much as latency.

Voice is becoming an interface.

Now it has to become one we can trust.