Automotive Speech-Recognition - Success Conditions Beyond Recognition Rates
archive: archived pipeline: cataloged verified
Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)
Summary
This paper argues that evaluating automotive speech-recognition systems requires looking beyond technical recognition rates to address usability factors such as mental workload, distraction, and learnability. The author posits that while speech technology is increasingly common in office and telephone contexts, automotive applications present unique challenges due to the coexistence of driving (primary task) and speech interaction (secondary task). Consequently, standard Human-Machine Interface (HMI) approaches are insufficient; specific automotive requirements, such as multimodal error recovery and facilities for user-motivated interrupts, are necessary to manage the increased workload and potential distraction caused by error-prone input modalities. The study draws on two user studies conducted in 1999 to illustrate these points. First, a field trial by Schweigert (1999) examined driver behavior during a secondary listening task across different traffic situations (highway, suburban, and inner city). Using eye-tracking technology, the study measured mental workload and distraction. Results indicated that mental load significantly altered visual scanning behavior, leading to reduced variability and longer fixation times on critical objects like the road and preceding cars. The analysis revealed that drivers employed specific strategies to shift attention between tasks to compensate for increased workload, with scanning activity decreasing differently depending on the complexity of the driving situation. This confirms that in-car speech systems must account for compensation effects in demanding traffic scenarios, including extended reaction times and syntax neglect. Second, the paper addresses learnability and error handling through an evaluation of a prototypical voice-dialing system by Niedermaier (1999). The authors utilized a structured error taxonomy to classify errors as either "system-based" or "user-based." Initial data showed a high frequency of errors related to timing ("spoken too late") and wording ("wrong initial command word"), stemming from user unfamiliarity with the dialogue flow. To address this, the dialogue design was modified to provide help text prompts if user input was delayed. This intervention reduced the targeted error types by 75%, although it inadvertently increased "spoken too early" errors as users responded more quickly to the new prompts. This demonstrates that errors serve as informative resources for optimization, provided a structured approach is used to prevent merely shifting error frequencies. The significance of this work lies in establishing that dialogue design is as critical as recognition accuracy for the success of in-car speech systems. The authors conclude that in-car speech recognition must be viewed as a scenario where the dialogue task is secondary to driving. They advocate for a structured, error-driven approach to HMI design and evaluation, suggesting that future systems should incorporate features like tailored help prompts and transparent dialogue states to support novice users. The paper calls for the development of valid parameters, beyond eye movements, to evaluate the compatibility of dialogue systems with various driving situations, ensuring that speech interfaces enhance rather than hinder traffic safety.
Provenance
The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.
| Stage | Outcome | Tool | Model | Prompt | Attempts | Completed |
|---|---|---|---|---|---|---|
| discover | success | Crossref | — | — | 1 | 2026-06-25 |
| archive | success | canonical_url | — | — | 1 | 2026-06-26 |
| extract | success | cached | — | — | 2 | 2026-06-26 |
| clean | success | clean | — | — | 1 | 2026-06-26 |
| chunk | success | chunk | — | — | 1 | 2026-06-26 |
| embed | success | embed | Qwen/Qwen3-Embedding-8B | — | 1 | 2026-06-26 |
| enrich | success | openalex | — | — | 1 | 2026-06-26 |
| promote | success | — | — | — | 1 | 2026-06-25 |
| summarize | success | llm | qwen3.6-27b-prismaquant | summ-v5 | 1 | 2026-06-26 |
| tag | success | vector_similarity | — | — | 6 | 2026-06-26 |
| verify | success | — | — | — | 1 | 2026-06-26 |
Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.
Topics
Ranked by relevance to this paper. Hover a topic for its definition.
Information type
What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).
- Applied Guidance: design guidelines
- Theoretical Contribution: conceptual framework