Speechify

Speechify Simba 3.2 has achieved top rankings on two independent text-to-speech benchmarks, highlighting its capabilities for real-time voice AI applications. "This is the underdog story for API providers," said Luke Oliff, Head of Developer Relations at Speechify. Read the full news: https://lnkd.in/diDyDMcZ #MachineLearning #AIInnovation #SpeechSynthesis #TechIntelPro

Five years ago, my Stanford University grad course project was fine-tuning Google's best text-to-speech model. This week, the model my team built from scratch at Speechify became the #1-ranked TTS model in the world. In 2021 I was a grad student planning to go down the systems track. Then I took Andrew Ng's ML course. The syllabus changed every single week because the field was moving too fast to pin down. That was enough to make me switch paths. That quarter I fine-tuned Tacotron2, Google's state-of-the-art TTS at the time. My TA's feedback is in the screenshot below: "This is fantastic work, Tyler." After the course ended, I decided to keep going. My work wasn't finished. I built the AI lab in Speechify with the goal of building the best text to speech model in the world from scratch. Today, we achieved our goal, 5 years later, with Simba 3.2. It's Speechify's own architecture, designed from the ground up over years of work. And as of this week it's ranked #1 in the world on Artificial Analysis' Speech Arena, ahead of every lab in the field. Google is #2. Yesterday Andrew Ng congratulated me on X. My old TA replied too. People see the leaderboard and think it happened fast. It didn't. It took five years to get here, and it started with a course project. – Written with Speechify Voice Typing

This week, Speechify's SIMBA 3.2 hits the No. 1 overall slot on Artificial Analysis overall global speech leaderboard. This is the widest leaderboard in the world for AI voice. It includes U.S. & Chinese models. The hyperscalers, the labs, and startups. Congratulations to Tyler Weitzman for spearheading this effort. Tyler started this research while at Stanford University five years ago. Since then, he has been working towards a single goal of making Speechify the top AI voice model provider in the world focusing on both quality & cost. I've been lucky to know Tyler for 15+ years as we were in the same high school dorm. While we may have been able to go faster just focused on quality, our consumer DNA and business model called for architecture decisions early on to make sure we could basically provide unlimited speech to our users for $139/year. That translates into what is today the SOTA TTS model at the best-in-class price, which is rare in AI research where better performance usually means higher costs. Excited for this next phase of Speechify, as others get to use our models to build their own amazing experiences and push the bounds of how AI can extend our work. Check out speechify.ai to get started. - Written with Speechify Voice Typing

Proud moment for the team Speechify's Simba 3.2 is now ranked #1 on the Artificial Analysis Speech Leaderboard. It’s been exciting to watch this come together from inside the company. Huge congrats to everyone who contributed to making this possible. 👏 Artificial Analysis leaderboard: https://lnkd.in/eDrEq7-w

Today, Speechify's SIMBA 3.2 took the No. 1 spot on the Artificial Analysis leaderboard for the top AI voice models in the world, above Google, ElevenLabs, OpenAI, and others. SIMBA 3.2 is also the most cost-efficient model on the entire leaderboard. Congrats to Tyler Weitzman & the entire AI team on reaching this milestone as the top ranked AI speech lab on multiple axes. To start using SIMBA 3.2 today, visit speechify.ai. - Written with Speechify Voice Typing

The SpeechSDK now supports Speechify's latest SIMBA models. Topping VoiceArena and currently tied for the top spot, I'm personally excited to use it in Jellypod, and to see all the other applications that depend on the SpeechSDK to try it out. It's been awesome to work with the team over there (shoutout Russell Sherrill) to bring this model to our open-source SDK. The idea behind the SpeechSDK was simple: text-to-speech audio models should compete on the basis of quality, cost, and features, not vendor lock-in. Every model today has very different API shapes, supporting different file types, sample rates, audio tags, etc. Developers shouldn't need to re-write their entire TTS generation stack from scratch when they want to try out a new model that might work better for them. The SpeechSDK makes swapping models a trivial single line change, while elevating the developer experience to enable seamless long-form generation, voice cloning, and multi-speaker conversations.

✨ WE'RE NUMBER ONE ✨ Speechify just blasted onto Voice Arena's US English leaderboard with our latest realtime TTS model, Simba 3.2, and now the defacto #1 streaming TTS model (technically tied with Cartesia Sonic-3.5 who are 7x more expensive 👀) https://lnkd.in/em3RAsGh #SpeechifyAI #TTS Cliff Weitzman Rohan Pavuluri Pankaj Agarwal Simeon Kostadinov Jason Kirschtel Kai Krause Tyler Weitzman

🚨 BREAKING: Simba 3.2 from Speechify is now the #1 streaming TTS model, statistically tied with Cartesia Sonic-3.5, on the Voice Arena US English 🇺🇸 leaderboard. It sits ahead of ElevenLabs v3 and OpenAI's gpt-4o-mini-tts. The economics seal it: $6 per 1M characters, nearly 7x below the model it shares the top spot with, at <100ms latency. The board is built on 23,234 blind, head-to-head listener votes. Every ranking on Voice Arena comes from blind pairwise A/B tests judged only by vetted native speakers. No self-reported benchmarks, no cherry-picked demos. Congratulations to the Speechify team 👏 Full US English board → https://lnkd.in/gJ8aHqrC

Speechify's SIMBA 3.2 is now the No. 1 ranked streaming AI voice model on Voice Arena, above ElevenLabs, OpenAI, xAI, and others. At $6 per 1M characters, it's also the lowest cost model on the leaderboard. That's 15x cheaper than ElevenLabs and 6x cheaper than Cartesia. After text-based LLMs, voice is one of the most competitive frontiers in AI. There's a golden trifecta developers care about: cost, latency, and quality. Most models force you to pick two. We've spent five years making architecture decisions that refuse that tradeoff. One advantage we've had: the largest consumer platform delivering AI voice in the world. Millions of real users, countless A/B tests, constant signal on what actually sounds good vs. what doesn't. All of that learning is baked into our SIMBA models. We built this for developers and companies who want SOTA quality at the best pricing on the market. Check it out at speechify.ai – Written with Speechify Voice Typing