Whisper-class models vs. the engine Lven Instant ships: a one-laptop measurement
We measured whisper-large-v3 and distil-large-v3 against the production engine Lven Instant ships, running on CPU, on one consumer laptop (Intel i5-10300H, 4 cores/8 threads; NVIDIA GTX 1650 4 GB) over 73 clean-speech clips. whisper-large-v3 was about one point more accurate on this set, and 7–40× slower depending on configuration. Realtime dictation needs a real-time factor well below 1; on this hardware only our engine delivers it.
Setup
Hardware: a mainstream consumer laptop — Intel i5-10300H (4 cores / 8 threads) with an NVIDIA GTX 1650 (4 GB VRAM). Test set: 73 clips of clean read English speech (the librispeech_asr_dummy set). Metrics: word error rate (WER), real-time factor (RTF — processing time divided by audio duration; below 1 means faster than realtime), and median per-clip latency.
Candidates: the shipping Lven Instant engine on CPU, against whisper-large-v3 (fp16 on the GTX 1650, and fp32 on CPU) and distil-large-v3.
Results
| Engine / configuration | WER | RTF | Median per clip |
|---|---|---|---|
| Lven Instant engine — CPU | 3.76% (full 73) | 0.166 | 0.93 s |
| whisper-large-v3 — fp16, GTX 1650 | 1.79% (paired first 24; Lven Instant: 2.79%) | 1.14 | 9.2 s (3.36 GB VRAM) |
| whisper-large-v3 — fp32, CPU | — | 2.46 | 27.6 s |
| distil-large-v3 — GTX 1650 | 4.11% (full 73) | 1.04 | — |
whisper-large-v3 earns its reputation for accuracy — roughly one point better on this clean-speech set. It also ran at RTF 1.14 on the GPU and 2.46 on CPU: slower than realtime in every configuration on this machine. The distilled variant was faster but scored worse than our engine on the full set (4.11% vs 3.76%).
Conclusion
For live dictation on mainstream consumer hardware, the constraint is not accuracy in isolation — it is accuracy at an RTF far enough below 1 that text lands as you speak. On this machine, our engine transcribes at RTF 0.166 with competitive accuracy; the Whisper-class models do not reach realtime at all. That is why Lven Instant ships this engine on its hot path, and why larger models remain interesting for server-side or opt-in accuracy modes rather than live typing.
Limitations.
- One laptop. Different CPUs and GPUs will shift every number; measure on your own hardware before deciding.
- Clean read English speech — noisy, accented, or conversational audio can reorder the accuracy results.
- Each engine was run in its practical deployment mode rather than a matched-precision ablation.
- The paired-WER comparison for whisper-large-v3 covers the first 24 clips only.
- ClassEve-run, first-party measurement — not an independent benchmark.