Whisper transcribes 70-year-olds more accurately than 20-year-olds, but voice-agent turn-taking lags 2-5x

Show HN: Whisper transcribes 70-year-olds more accurately than 20-year-olds

This open-source project tested Whisper across age groups and found that speakers over 70 had lower word error rates than 20-year-olds. The real bottleneck is voice-agent turn-taking: older adults pause and hesitate more, causing 2-5x longer end-to-end latency. The post doesn't disclose sample size or recording conditions, so take the exact numbers with a grain of salt, but the direction matters—interaction pacing, not just ASR accuracy, is the hard problem for older users.

Why it matters: Counterintuitive finding with concrete numbers reframes voice-agent accessibility from recognition accuracy to interaction latency — directly useful for conversational AI teams. Score capped because sample size and recording conditions aren't disclosed, so the numbers are dire...