Modulate

We're hiring. 3 roles, 1 mission: make voice a force for good online.

→ Backend Engineer
→ Staff Backend Engineer
→ Senior ML Engineer

Somerville, MA. Full time. Hybrid.

Most developers don't switch speech-to-text providers for a small win. They switch when one option is obviously, undeniably better 🏅

That's what happened here:
⭐ 5/5 "Outstanding Accuracy, Speed, and Value"

Simple API. Clear docs. Integration that takes minutes, not days.
Start building.👇

Speech recognition = audio in, text out. That's the easy version.

Our CEO and Co-founder @whuffman on why real understanding means reading accent, tone, emotion, and cadence too - not just the words.

We're hiring for people who think this way.

See open roles: hubs.ly/Q04xDKp90

#hiring

Winnowing nuance from noise
Modulate runs over 100 voice-native models to discern emotion, tone, intent, and language, and detect synthetic and deepfake voices (ranked #1 on 🤗).
We just led their seed round; news today: techcrunch.com/2026/09/28/mod…
Product: modulate.ai/velma

🚨 A startup just hit #1 on @HuggingFace for deepfake audio detection.

98.9% accuracy. 30–1000× cheaper than every competitor. Nobody is talking about this.

It's called Velma by @modulate_ai and it changes everything about voice security.

🚨 BREAKING: Transcription just got 90x cheaper.

One of the biggest hidden costs for AI startups is Speech-to-Text.

Velma-2 by @modulate_ai just launched: $0.03 per audio hour transcribe.

Here’s why this changes everything for AI founders:

I've built two voice pipelines for two different companies.

They both look like this:

Audio → STT → Clean transcript → NLP → Classify → Act

This works, but there's still a problem I can't solve.

Every time I convert audio to text, I'm keeping the words but throwing away the meaning. Tone, hesitation, sarcasm, and stress are all gone. I have the text, but miss its soul.

The folks at @modulate_ai reached out and showed me how to solve this.

Velma is the voice model that's been running inside Call of Duty and GTA Online to catch toxicity in real time.

This model skips the transcript entirely and works directly on the raw audio. This allows the model to take into account the "invisible clues" other models miss.

It can detect up to 150 invisible clues that none else does!

You can access Velma through an API, and it's ~10x cheaper than pushing audio through an LLM.

If you want to give it a try, use this link to get 1,000 free credits:

modulate.ai/api/velma?utm_…

Thanks to the team for partnering with me on this post.

The weird thing about voice AI today?

It doesn't actually understand voice.

audio → transcript → LLM

And right there, before your model sees a single token, tone, stress, hesitation, sarcasm, and intent are already gone.

You've been reasoning about conversations from their shadows.

This startup just solved this problem.

@modulate_ai built Velma to fix this.

The ONLY audio-native model of it's kind
- trained on 550M hours of raw audio
150+ behaviors detected in real time - straight from the audio, no transcript step.

Velma is already running inside Call of Duty and GTA Online.

Now it's open to every developer.

Get 1,000 free credits in the comment below 👇

8. @modulate_ai announced $25M in new funding. Its Velma platform analyzes audio directly for signals including tone, synthetic speech and conversational behavior. Agent supervision is one application.

Founders get told to ignore the doubters. So do early employees.
Our co-founder Mike: on knowing when to trust your gut vs. when the "obvious" advice is actually right.

We're looking for people who can tell the difference 🧠 : hubs.ly/Q04xDKsr0

#hiring

What does a small IT services team actually need from a speech-to-text API?

Speed. Accuracy. A price that doesn't eat the margin.

One G2 reviewer found all three with Modulate Transcribe:
⭐ 5/5 "Modern, Cost-Efficient STT with a Standout Feature Set"

They are now using it to power AI voice chat for their own clients - at a price point that lets them deliver more value downstream.

Start building with our API today 👇

Last week, we joined the developer community at @WeAreDevs in San Jose 🙌

Great conversations, new connections, and plenty of discussions around what’s next for AI, voice, and the developer ecosystem.

"A lot of people assumed we were on our way to being acquired. That's never been the mission."

@modulate_ai, Co-founder Mike Pappas on why the company was built to go the distance - not to get bought.

👇

We just raised $25M in new funding led by Future Ventures (@VenturesVc), with @HyperplaneVC and @lakestarvc - bringing Modulate’s total funding to $60 million.

Building the next phase of audio-native AI: more models, more developers, more of the stack that actually understands voice.

Where we stand today:

- Ranked #1 on Hugging Face for deepfake detection and transcription
- 600M+ hours of audio analyzed
- 10M+ hours analyzed every month

A foundational layer of the AI stack is getting more powerful. More soon.

Most fraud tools are reading the wrong thing. 👀

They monitor transactions. They flag metadata. They analyze the transcript after the call ends.

All useful, BUT none of it catches fraud while it's happening, because the tells aren't in the words. They're in the voice.

Here's what a transcript misses on every fraud call:
1️⃣ Scripted or rehearsed speech patterns
2️⃣ Emotional incongruence - calm under pressure, urgency without stress
3️⃣ Hesitation, contradictions, and recall gaps
4️⃣ Voice mismatch against the account holder
5️⃣ Synthetic or AI-generated voice indicators

Modulate's Velma is built to catch all five live: flagging deceptive intent to your agents before the call completes, not after the damage is done.

Account takeover fraud cost US adults $15.6B in 2024, up 23% year over year. 📈

And Velma's deepfake detector generates fewer than half the errors of the next-best model on @huggingface independent leaderboard.

Stop reading the transcript. Start listening to the call: hubs.ly/Q04xdbt90

This team built the first AI models that understand voice - not just what's said, but how it's said.

In a few hours, we're sharing what's next. 👀

We’re hiring a Senior Machine Learning Engineer at Modulate 🔊

Voice is one of the richest sources of information we have - and we’re building models that can actually understand it.

This role will work across research and engineering to design, train, evaluate, and deploy ML models powering our conversational voice intelligence platform.

You’ll be working on problems spanning model quality, robustness, scalability, and production deployment - and helping push forward what machines can understand from human conversation.

If you have experience building and shipping production-grade ML systems, strong Python and PyTorch skills, and want to work on some genuinely interesting problems in voice AI, we’d love to hear from you! 🙌 👇

#hiring

Clinical audiology transcription needs more than words on a page - it needs context.

One G2 reviewer built an AI Scribe SaaS on Velma 2 for exactly that reason:

⭐ 5/5 "Accurate Transcriptions with Emotional Context at a Fantastic Price"

Live diarization.
Emotional context per utterance.
Accuracy that holds up in regulated, high-stakes use cases.

That's Velma 2, built for developers who need transcription that understands more than just what was said 👀

See what you can build 👇

🔉 We’re hiring a Backend Engineer at Modulate.

Help build and scale the Developer and Enterprise APIs powering our voice AI models. Ship code that puts ML audio models into production - not just papers.

Apply here: hubs.ly/Q04vH6HT0