@ASHLEYREESE_AI
Account based inPakistan
About this account
- Account based in
- Pakistan
- Connected via
- Pakistan Android App
Account-level information from X, not a live location or the device used for a specific post.
AI creator exploring the future of Tech. • Innovation & Digital Culture
Joined June 2026
- Tweets247
- Following10
- Followers21.6K
- Likes285
Ashley Reese | AI retweeted
I used to spend hours trying to make AI voices sound real.
Now I just tell @FishAudio to "say it like you are super happy" and it works.
Drama 3 is crazy good.
🤝 Paid partnership
$52M seed round. 15M+ users. But this is what we're most proud of yet.
Introducing Drama 3: the most emotive and controllable TTS model in the world
And yes, it's open-weight.
You can now control how even a single word should sound, change its tone, and shift your voice mid-sentence.
Uncle Roger was impressed too.
Try it here:
s.fish.audio/rqh0l8
Comment “DRAMA” within the next 48 hours for 1 month free.
If you're an enterprise your first month is on us:
s.fish.audio/qz9r0t
crazy how agents mastered email, calendars, cards and carts this year but still can't tell you what's on your desk. Collov made that the whole point with a personal agent with eyes. fair gap to chase.
Ashley Reese | AI retweeted
Muse gave your agent a computer.
Instinct gave it your phone number.
We gave it eyes.
Introducing NewEyes AI — your personal agent with eyes.
> Glasses see it.
> Earphones say it.
> It remembers where you left off.
> "Good morning. Continue on page 212."
Your world is the context.
Ashley Reese | AI retweeted
$52M seed round. 15M+ users. But this is what we're most proud of yet.
Introducing Drama 3: the most emotive and controllable TTS model in the world
And yes, it's open-weight.
You can now control how even a single word should sound, change its tone, and shift your voice mid-sentence.
Uncle Roger was impressed too.
Try it here:
s.fish.audio/rqh0l8
Comment “DRAMA” within the next 48 hours for 1 month free.
If you're an enterprise your first month is on us:
s.fish.audio/qz9r0t
Ashley Reese | AI retweeted
Nice that they opened up the model for research testing too.
Might run it on a few of my own clips this weekend and see if the numbers hold up.
Insights from training on Monsoon #1: same model, new data
We took Whisper Medium, an open 769M-parameter model, and changed nothing about it. Same architecture, standard fine-tuning recipe. The only new thing was the data: about 3,500 hours per language of real, unscripted conversations from Monsoon.
Word error rate on FLEURS, before → after:
Bengali 85.27% → 7.65%
Gujarati 75.17% → 6.63%
Marathi 63.05% → 9.90%
Tamil 37.49% → 14.35%
Telugu 85.48% → 9.95%
Average: 69.3% → 9.7%
At Voice Arena we don't want to make datasets, we want to solve problems. Solving speech recognition in the long tail of languages is a data problem. That is why Voice Arena built the Monsoon ASR corpus.
Results and Model API for research testing only: voicearena.com/datasets#resu…
Ashley Reese | AI retweeted
A fire took Craig Veroni's home, his family's belongings, and his place to film. For a realtor whose leads came from YouTube, that was the business too.
He made a personal avatar and posted every day. In 10 months his Instagram grew 4x.
Read more: heygen.com/customer-stories/…
Ashley Reese | AI retweeted
Introducing SubQ Snap, the first long-context decision model for real enterprise work.
Snap is the only decision model or LLM where compute per token stays constant as context grows.
• 384K tokens of multimodal context, 12x larger than Jev.
• 61.7 on Decision Index v0.2.1 vs Jev’s 57.9, outperforming on 13 of 16 enterprise benchmarks.
•Processes 100K tokens and returns a decision in under one second.
•$0.05/MM input tokens. Output is free.
Available today to anyone.
The first 10,000 sign-ups get 100M free input tokens.
We’ll likely hit capacity fast and will add users as we add compute. Compute offers welcome 🙂.
Latency will be variable as we tune the platform to this new model and category of users.
Model speeds are 80ms for minimal context, 250ms for 32k, 660ms for 100k, and 2.5 seconds for 384k.
Visit: subq.ai/
Ashley Reese | AI retweeted
If the subject consistency is as good as it sounds, this could be handy for building a visual identity across several images.
Check it out: flova.ai/
The real test is whether you can make a small change without messing up everything else in the image.
See how Nano Banana 2.1 handles it: flova.ai/
Want to see how your most detailed video prompts translate into moving images? Vidu Q4 is now available on @Flovaai, bringing together cinematic camera movement, high-resolution visuals, and a focus on prompt accuracy: flova.ai/
Autocomplete is dead.
Sabi raised $50M for a cap that reads your mind and types what you want before you even start.
we've raised $50M from Khosla Ventures, Accel, Initialized, Kevin Weil, DST Global
to build the world's most wearable BCI cap at Sabi
read more here:
forbes.com/sites/rashishriva…
Ashley Reese | AI retweeted
Here’s CRIS-0 in a real home: handling household tasks, checking what happened, and adjusting when the scene changes.
Ashley Reese | AI retweeted
we've raised $50M from Khosla Ventures, Accel, Initialized, Kevin Weil, DST Global
to build the world's most wearable BCI cap at Sabi
read more here:
forbes.com/sites/rashishriva…
Ashley Reese | AI retweeted
Most data vendors ask you to trust them. @voicearena_ai lets you test them.
Monsoon launched with its results public (Whisper Medium's Bengali WER down from 85% to 7.65%) and an API open for labs to run their own checks.
80+ licensing requests in the first week says the approach works.
Crazy week for Voice Arena at Interspeech in Sydney.
80+ organisations have asked to license Monsoon ASR corpus since we launched it seven days ago.
The most common reason why labs are interested: Monsoon promises results.
When we decided to build datasets at Voice Arena, we set one rule. Either the dataset promises results, or we don't build it.
Monsoon promises results. On Bengali FLEURS, fine-tuning Whisper Medium on Monsoon took its LLM word error rate from 85.27% down to 7.65%. The other thing labs like is that we give them access to our model API. They can test it on their own internal benchmarks and see for themselves whether it will improve their models.
You can see the detailed results and get API access here: voicearena.com/datasets/mons…
Monsoon is 100,000 hours across 50 languages from around the world, and most of them are long-tail languages. 100 languages by February, 1,000 by the end of 2027.
It's only the first dataset Voice Arena has launched. We are excited to keep working on the key problems that get us closer to the dream of machines that talk like humans.
Ashley Reese | AI retweeted
80+ licensing requests in a week is unusual for a dataset.
The reason is simple: Monsoon shipped with proof. @voicearena_ai fine-tuned Whisper Medium and took Bengali WER from 85% to under 8%, then opened the test API so labs could verify it on their own benchmarks. No promises, just measured results.
Crazy week for Voice Arena at Interspeech in Sydney.
80+ organisations have asked to license Monsoon ASR corpus since we launched it seven days ago.
The most common reason why labs are interested: Monsoon promises results.
When we decided to build datasets at Voice Arena, we set one rule. Either the dataset promises results, or we don't build it.
Monsoon promises results. On Bengali FLEURS, fine-tuning Whisper Medium on Monsoon took its LLM word error rate from 85.27% down to 7.65%. The other thing labs like is that we give them access to our model API. They can test it on their own internal benchmarks and see for themselves whether it will improve their models.
You can see the detailed results and get API access here: voicearena.com/datasets/mons…
Monsoon is 100,000 hours across 50 languages from around the world, and most of them are long-tail languages. 100 languages by February, 1,000 by the end of 2027.
It's only the first dataset Voice Arena has launched. We are excited to keep working on the key problems that get us closer to the dream of machines that talk like humans.
Ashley Reese | AI retweeted
One of the biggest names in real estate now has a daily video series he doesn't film.
Today we're launching it with @RyanSerhant, delivered by his official avatar. Sales, business, and building your brand. Every day.
The full story in Variety: variety.com/2026/biz/news/ow…
