The Open ASR Leaderboard incorporated private evaluation splits to improve accent diversity and mitigate benchmaxxing. Through a partnership with Liva AI, the platform added spontaneous conversational audio covering both Egyptian and Nigerian English speakers. These files capture natural user behaviors, complete with verbatim disfluencies, casing, and punctuation. The integration builds on earlier held-out datasets sourced from Appen, DataOceanAI, and Voice Arena.
Key points
Incorporating held-out test splits helps ensure speech recognition benchmarks remain trus…: Incorporating held-out test splits helps ensure speech recognition benchmarks remain trustworthy and resistant to artificial score inflation.
Capturing audio directly through consumer software products provides more natural behavio…: Capturing audio directly through consumer software products provides more natural behavioral data than traditional scripted or task-based collection techniques.
Expanding regional accent coverage addresses specific performance blind spots across dive…: Expanding regional accent coverage addresses specific performance blind spots across diverse English dialects in automatic speech recognition systems.
Sources checked by our analysis pipeline: 1
RW-Voice EQ ASR
Explore and compare English speech‑recognition model performance
A few months ago, we added private data to the Open ASR Leaderboard: 11 splits from Appen and DataOceanAI covering scripted and conversational speech from five accents (blog). The community response was great, with many highlighting the importance of held-out sets for trustworthy evaluations. Since then, we added one more private split for Indian-accented English with Voice Arena (blog).
We’re excited to announce a new collaboration with Liva AI to add even more private data with two new accents: Egyptian 🇪🇬 and Nigerian 🇳🇬.
Liva AI Conversational EG | Egyptian | 1.34 | Conversational, spontaneous | Punctuated, cased, with disfluencies. |
Liva AI Conversational NG | Nigerian | 1.55 | Conversational, spontaneous | Punctuated, cased, with disfluencies. |
This is part of our continued effort to improve representation in ASR evals (more accents and more languages) and to improve trust (new private data to minimize benchmaxxing, and new data providers). We’re excited to see the community continuing to…
The Y Combinator-backed robotics platform is expanding its technical workforce to advance an embodied artificial intelligence system for industrial settings. Backed by top investors and partnering with manufacturing leaders like ABB, the company is bridging generative models and physical factory workflows. This recruitment effort supports the deployment of systems that let factory machines acquire complex skills through everyday spoken instructions.
Congressional lawmakers are laying the groundwork for a sweeping legislative push focused on emerging technologies. The upcoming measures aim to address critical governance and oversight gaps in the rapidly developing sector. This strategy highlights an aggressive effort to establish federal guardrails as soon as legislative power shifts.
Lawmakers are responding to voter anxieties surrounding unregulated technology risks and the heavy power consumption of computing infrastructure. The impending legislative push targets tech industry accountability, independent safety testing, and protection for household utility rates. These regulatory efforts have already rattled investors, helping trigger a notable market downturn across technology stocks.