Benchmarking Commercial ASR Systems on Code-Switching Speech: Arabic, Persian, and German
arXiv cs.CL·Sajjad Abdoli (MAD), Ghassan Al-Sumaidaee (MAD), Clayton W. Taylor (MAD), Ahmad (MAD), ElShiekh, Ahmed Rashad
17h agoQuick Take
The study benchmarks commercial ASR systems on code-switching involving Arabic, Persian, and German languages.
Key Points
- Evaluates five ASR providers across four language pairs.
- Introduces a cost-effective two-stage dataset selection pipeline.
- Finds ElevenLabs Scribe v2 leads in WER and BERTScore.
📖 Reader Mode
~2 min readAbstract:Code-switching -- the natural alternation between two languages within a single utterance -- represents one of the most challenging and under-studied conditions for automatic speech recognition (ASR). Existing commercial ASR benchmarks predominantly evaluate clean, monolingual audio and report a single Word Error Rate (WER) figure that tells practitioners little about real-world multilingual performance. We present a benchmark evaluating five commercial ASR providers across four language pairs: Egyptian Arabic--English, Saudi Arabic (Najdi/Hijazi)--English, Persian (Farsi)--English, and German--English. Each dataset comprises 300 samples selected by a two-stage pipeline: a heuristic filter scoring transcripts on five structural code-switching signals, followed by a GPT-4o and Gemini 1.5 Pro ensemble scoring candidates across six linguistic dimensions. This pipeline reduces LLM scoring costs by approximately 91\% relative to exhaustive scoring. We evaluate the systems on both WER and BERTScore, arguing that BERTScore is a more reliable metric for Arabic and Persian pairs where transliteration variance causes WER to penalise semantically correct transcriptions. ElevenLabs Scribe v2 achieves the lowest WER across all four language pairs (13.2% overall; 13.1% on Egyptian Arabic) and leads on BERTScore (0.936 overall). We further demonstrate that difficulty-stratified analysis reveals performance gaps masked by aggregate averages, and that BERT embedding projections confirm semantic proximity between reference and hypothesis despite surface-level script differences. The benchmarking dataset is publicly available at this https URL.
| Subjects: | Computation and Language (cs.CL); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2605.19069 [cs.CL] |
| (or arXiv:2605.19069v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2605.19069 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Sajjad Abdoli [view email]
[v1]
Mon, 18 May 2026 19:50:44 UTC (748 KB)
— Originally published at arxiv.org
More from arXiv cs.CL
See more →arXiv cs.CL·Leyao Wang, Yanan He, Peng Chen, Asaf Yehudai, Yixin Liu, Rex Ying, Michal Shmueli-Scheuer, Arman Cohan
17h ago FeaturedOriginal
Time to REFLECT: Can We Trust LLM Judges for Evidence-based Research Agents?
AI Summary
The reliability of LLM judges for evaluating deep research agents is critically assessed using the REFLECT benchmark.