Hi HN, I’m Kajo, co-founder of KugelAudio (YC P26).
We build voice models focusing on customer support. One of the main issues our customers are having right now is calls that drift into endless spelling loops and lots of false interruptions, especially in non-English languages.
A classic example would be the name “Philipp” or “Philip”. The customer says their name on the phone and the bot either doesn't know about the ambiguity or lets users spell everything letter by letter, which is extremely annoying and prone to error or interruption mid-spelling.
This is why we built Kugel Ear. When it isn't sure, it returns other spellings with probabilities, so the bot can ask one targeted question: “Philipp with two P’s?” Kugel Ear also comes with integrated context-aware EOT (end-of-turn) detection so the voicebot knows the user is mid-spelling and doesn't interrupt.
We score 1st on LiveKit EOT bench, 1st on AppTek’s call-center dialogues, 1st on calls over phone lines, and 2nd on VoiceCodeBench behind OpenAI, with a median latency of 72 ms from end of speech to final transcript. We don’t win everything: on Earnings22 (long earnings calls) ElevenLabs Scribe beats us (4.0% vs 4.4% WER), and on VoiceCodeBench OpenAI writes code and numbers slightly more accurately (83.1% vs 81.9%).
False cut-off rate
(pauses taken as end of turn at 300 ms average wait, lower is better, next best = best published model):
I’d love to hear where it breaks for you, especially calls where it still cuts someone off or misses a spelling. I’ll be in the thread all day answering questions.
kajokr•1h ago
A classic example would be the name “Philipp” or “Philip”. The customer says their name on the phone and the bot either doesn't know about the ambiguity or lets users spell everything letter by letter, which is extremely annoying and prone to error or interruption mid-spelling.
This is why we built Kugel Ear. When it isn't sure, it returns other spellings with probabilities, so the bot can ask one targeted question: “Philipp with two P’s?” Kugel Ear also comes with integrated context-aware EOT (end-of-turn) detection so the voicebot knows the user is mid-spelling and doesn't interrupt.
We score 1st on LiveKit EOT bench, 1st on AppTek’s call-center dialogues, 1st on calls over phone lines, and 2nd on VoiceCodeBench behind OpenAI, with a median latency of 72 ms from end of speech to final transcript. We don’t win everything: on Earnings22 (long earnings calls) ElevenLabs Scribe beats us (4.0% vs 4.4% WER), and on VoiceCodeBench OpenAI writes code and numbers slightly more accurately (83.1% vs 81.9%).
False cut-off rate (pauses taken as end of turn at 300 ms average wait, lower is better, next best = best published model):
German: 5.9% (next best: 11.8%)
Spanish: 3.8% (next best: 15.5%)
French: 5.0% (next best: 12.7%)
Italian: 5.8% (next best: 16.9%)
Dutch: 4.7% (next best: 12.5%)
Portuguese: 5.6% (next best: 19.3%)
English: 7.4% (next best: 9.9%)
It costs €0.48 per audio hour (about $0.55). You can try it out at https://kugelaudio.com/kugel-ear-1
I’d love to hear where it breaks for you, especially calls where it still cuts someone off or misses a spelling. I’ll be in the thread all day answering questions.
Datasets:
- Support calls: 280 turns from AppTek’s call-center dialogues, 20 per accent across 14 accents, median WER. https://huggingface.co/datasets/apptek-com/apptek_callcenter... - Phone line: the same 280 turns through a simulated G.711 line (8 kHz, mu-law, noise, dropouts). - German names/addresses: 800 synthetic German calls we generated, half through the same phone line. - Codes and numbers: VoiceCodeBench, scored with deterministic exact match instead of its LLM verifier. https://huggingface.co/datasets/besimple-ai/voice-code-bench - Earnings22: Artificial Analysis’ cleaned version. https://huggingface.co/datasets/ArtificialAnalysis/Earnings2... - End of turn: LiveKit’s eot-bench. https://github.com/livekit/eot-bench and https://huggingface.co/datasets/livekit/eot-bench-data