Preprint
Measuring the Accuracy of Automatic Speech Recognition Solutions
Korbinian Kuhn et al., arXiv (cs.CL), preprint, 2024
Read the originalRead on October 16, 2026
- What it is
- A preprint that measured the accuracy of eleven common automatic speech recognition services on recordings of university lectures, including streaming services used for live events.
- How much weight it carries
- A preprint, so it has not passed peer review at the time of access. It tests one kind of content, university lectures, so its results describe that material rather than every recording.
Facts taken from this source
- The authors conclude that, despite recent improvements, common ASR services lack reliability in accuracy.
“Our study shows that despite the recent improvements of ASR, common services lack reliability in accuracy” (Abstract)
- The study found significantly lower quality for streaming speech recognition, the kind used for live events.
“We also measured a significant lower quality for streaming ASR, which is used for live events” (Abstract)
- The study measured eleven common automatic speech recognition services on recordings of university lectures.
“We measured the performance of eleven common ASR services with recordings of Higher Education lectures” (Abstract)
Pages that rest on this source
- Speech Recognition: What It Is and How Well It WorksTopic
- Best Dictation Software for Windows: Free and PaidGuide
- Stenography and Shorthand: Writing at 225 WPMGuide
- Is Windows dictation any good?Question
- How Accurate Is Speech Recognition? The Real NumbersGuide
- Real-Time Speech Recognition: Streaming and LatencyGuide
- Zoom, Teams and Google Meet TranscriptionGuide
- Do speech to text glasses work for deaf people?Question