Multilingual meeting AI, measured transparently.
Beesiness transcribes multilingual business conversations at 8.9% WER on our internal evaluation set — on par with the best published provider figures we've seen, and roughly three times more accurate than the open-source baseline. The sections below show the numbers, the methodology, and five real code-switching examples across English, Spanish, German, and French.
Multilingual ASR Word Error Rate (lower is better)
| Model | WER % | Note |
|---|---|---|
Beesiness production pipelineours | 8.9% | Averaged across EN, ES, DE, FR, PT, AR business-meeting samples |
Industry baseline (zero-shot, multilingual) | 25.4% | A public, open-source baseline on the same evaluation set |
Beesiness next-gen (Q4 2026 target)target | 6.5% | In-house fine-tuning target — not yet released |
Methodology.
- Evaluation set: 120 minutes of held-out, consent-recorded business meetings across EN, ES, DE, FR, PT, AR — with code-switching in ~30% of clips. Sales, product, engineering, and education domains are represented. Shared under NDA with qualified enterprise buyers running a procurement review; not yet public.
- Reference transcripts: two human annotators, with disagreements resolved by a third reviewer. Standard NIST WER scoring; case-insensitive, punctuation removed.
- Baseline: an open-source ASR evaluated on the same audio (zero-shot, large multilingual model). We re-run the comparison whenever we ship a pipeline update.
- Independent replication: contact info@beesiness.com for the evaluation set under NDA. We're working toward a public release that doesn't expose customer audio.
Code-switching · five real meeting examples
Modern enterprise speech mixes English business jargon into any sentence frame — 'runway', 'churn', 'pipeline', 'rollback', 'SOC 2'. Most ASR + summarization tools "translate" these and destroy the meaning. Beesiness keeps them verbatim across the language switch.
"We need to extend the runway this quarter — Series A is off the table until churn drops below 3%."
✓ Beesiness
Beesiness keeps 'runway', 'churn', and 'Series A' verbatim. The action item is captured as: "Extend Q4 runway plan — bring churn under 3% before Series A pitch".
✗ Naive translation
Naive translation pipelines turn 'runway' → 'airport runway' and 'churn' → 'butter churn', destroying the business meaning.
"Necesitamos actualizar el deal stage en HubSpot — hay un agujero serio en el pipeline este mes."
✓ Beesiness
It recognizes HubSpot as a product name and keeps 'deal stage' and 'pipeline' in English even mid-Spanish-sentence. Action item: "Cerrar el gap en el HubSpot pipeline — este mes".
✗ Naive translation
Naive: HubSpot gets garbled by transliteration, 'pipeline' becomes 'tubería' (literal plumbing), and 'deal stage' is translated word for word.
"Vor dem Deploy in Production müssen die E2E-Tests grün sein — letztes Mal mussten wir einen rollback machen."
✓ Beesiness
The decision line is captured as: "No production deploy without green E2E tests — rollback risk from last time". The German frame stays intact while the English DevOps terms are preserved.
✗ Naive translation
Naive: 'rollback' → 'Rückwärtsbewegung', losing its deploy-pipeline-specific meaning.
"On va faire un showcase du sprint demo à la product team, et on iterate sur le feedback tout de suite."
✓ Beesiness
Action item: "Sprint demo showcase for product team — iterate on feedback the same day". Five English business terms are preserved inside the French sentence frame.
✗ Naive translation
Naive: 'showcase', 'feedback', and 'iterate' all get translated into clumsy local equivalents, losing the product-management nuance.
"To close the enterprise deal we need SOC 2 Type II — let's accelerate the audit."
✓ Beesiness
Action item: "Accelerate SOC 2 Type II audit to close enterprise deal". Compliance terms are preserved verbatim across the speaker switch.
✗ Naive translation
Naive: when non-English context appears, 'enterprise', 'audit', and 'Type II' get translated — and compliance reviewers reject the summary.
See it in your own meeting
60 minutes free, no credit card. The bot joins your next Meet, Zoom, or Teams call and you read the benchmark numbers from your own meeting.