Hyeongju Kim

CEO @ Out of Set

[Out of Set @outofset]

Airy TTS is here.
> $2 per 1M characters, up to 96% cheaper
> 0.2s end-to-end latency, up to 20x faster

Most TTS APIs focus on time-to-first-audio.
They're already fast there.

But end-to-end latency is a different story.
You're still waiting for the full response.

Airy TTS solves that.

You get a complete 27s audio clip back in just 0.2s.
Up to 20x faster than other APIs.

Try it now:
https://t.co/nvvxFi3uor

[Hyeongju Kim @anlgboy]

@Qalqi Sure. We're working on supporting more languages.

[Hyeongju Kim @anlgboy]

Most TTS APIs focus on time-to-first audio.
They're fast enough there.

But let's talk about end-to-end latency.
You're still waiting for the full response to finish.

Airy TTS solves that.

You get a complete 27s audio clip back in just 0.2s.
Up to 20x faster than other APIs.

You can test here:
https://t.co/qKHEAEGDVM

[Hyeongju Kim @anlgboy]

@AhmedReza Pricing is at the bottom of https://t.co/1ukZUaGzV4. We're not HIPAA-ready yet, and we'll update this if that changes.

[Hyeongju Kim @anlgboy]

@VictorZhangHC We're working on getting listed in external benchmarks, so quality will be verifiable soon.

[Hyeongju Kim @anlgboy]

@killswiitchh Or you may explore them here: https://t.co/kYiNyrbS98

[Hyeongju Kim @anlgboy]

@killswiitchh We provide 100+ voices by default. You can check here: https://t.co/ikvtdUkq3Y

[Hyeongju Kim @anlgboy]

@_PedroPizarro_ Yes, e2e latency varies by many factors like location and network conditions. The same applies to TTFA claims. Our point is that since our system is architected around e2e latency at the model and serving level, relative ordering should generally hold.

[Hyeongju Kim @anlgboy]

@JustLingonberry It’s around $2 - $5 per 1M characters, which is much cheaper than other providers (e.g., ElevenLabs Flash costs $50 per 1M chars).

[Hyeongju Kim @anlgboy]

@axgtes It currently supports English and Korean.

[Hyeongju Kim @anlgboy]

@ItsNash0 They're generated live. The demos call our TTS API directly, and the audio you hear is generated and returned in real time. Nothing is cached or simulated.

[Hyeongju Kim @anlgboy]

Airy TTS is released!
> $2 per 1M characters, up to 96% cheaper
> 0.2s end-to-end latency, up to 20x faster

It's live here:
https://t.co/uTLT7ZvcdT

[Hyeongju Kim @anlgboy]

@ItsNash0 Thank you! Hope you enjoy trying it out.

[Hyeongju Kim @anlgboy]

https://t.co/uTLT7ZvcdT

[Hyeongju Kim @anlgboy]

Airy offers the easiest way to create voice content.

You won't pay to use basic AI features.
You won't waste time waiting for a response.
You won't get lost in complicated UX.

Supported languages: 한국어, English

[Victor M @victormustar]

🚀 Supertonic 2 TTS is already here!

- Runs entirely on-device
- Generates 1 sec of audio in 0.006 sec
- Super impressive quality!
- 66M params, no cloud, no API calls
- 5 languages: EN, KO, ES, PT, FR

⬇️ Demo available on Hugging Face

[Hyeongju Kim @anlgboy]

Demo: https://t.co/iQWfyz8kDI

Model: https://t.co/YNArlH4LKa

Code: https://t.co/If1NtviG0r

[Hyeongju Kim @anlgboy]

Supertonic 2 is officially out!
└ 66M params · 5 languages · On-device TTS
└ Commercial use? Yes.

👇Supported Languages
한국어 · Español · Français · Português · English

[Supertone @Supertone_ai]

Read Aloud, trusted by more than 6 million users, has added Supertonic as a new voice option to its text-to-speech Chrome extension. 📢

The Read Aloud team praised Supertonic for its ability to deliver “amazing voices that can run on commodity devices,” and shared expectations for the voices’ adoption across the platform.

Try Supertonic on Read Aloud!
⚡ Chrome extension: https://t.co/AP2GeogOGm
⚡ MS Edge: https://t.co/BFRda7kLFJ
⚡ TTS Tool: https://t.co/yhQfO1YFJw

As a platform widely used in schools, Read Aloud’s adoption of Supertonic is expected to lower the barrier to high-quality text-to-speech without specialized hardware, as Supertonic is designed to deliver strong voice quality directly on-device. 📚

We’re grateful for the trust and feedback, and we’ll continue focusing on making voice AI more accessible.

[Rishi @notnotrishi]

Supertonic by @Supertone_ai is one hell of a TTS model 👏

Low memory footprint (just 66M params), instant generation, runs literally on any device -- and the best part is it can handle financial/technical units, date times, phone numbers etc. without any extra work.

(ignore the slight echo from my laptop, the model output is pretty solid)