Dialogist — Research https://dialogist.cc/en/research/ Research notes from the Dialogist team: latency, turn-taking, speech recognition and synthesis, measured on real calls and explained step by step. en Wed, 11 Nov 2026 00:00:00 GMT Voicecore: Rust media backend, road to infinite scalability with no idle GPUs https://dialogist.cc/en/research/voicecore-rust-media-core/ https://dialogist.cc/en/research/voicecore-rust-media-core/ Wed, 11 Nov 2026 00:00:00 GMT The Python media path used up the processor of one bot process at about 96 conversations. Voicecore, written in Rust, carries 224 with the GPU out of the way. Research Capacity Load One Server Unit: 100 simultaneous calls, load-tested to 176% of rated capacity https://dialogist.cc/en/research/server-unit-under-load/ https://dialogist.cc/en/research/server-unit-under-load/ Wed, 28 Oct 2026 00:00:00 GMT We loaded one Server Unit from a single call to 176% of its rated 100 calls. Replies held the Standard grade up to rated capacity, and every call connected. Research Load Capacity Oasis Speech 2: recognition in 48 languages, before and after https://dialogist.cc/en/research/oasis-speech-2/ https://dialogist.cc/en/research/oasis-speech-2/ Mon, 26 Oct 2026 00:00:00 GMT Oasis Speech 2 lowers the error rate on short replies by 44% across the 48 languages it recognizes, and all 48 improve. Research Speech recognition Languages Listening while talking, and what the agent does with the caller's open channel https://dialogist.cc/en/research/listening-while-talking/ https://dialogist.cc/en/research/listening-while-talking/ Sat, 24 Oct 2026 00:00:00 GMT How a Dialogist agent keeps listening while it speaks, starts its reply inside the pause and tells an interruption from a cough. Research Turn-taking The shortest answers are the hardest to recognize, and the most common on a call https://dialogist.cc/en/research/short-replies/ https://dialogist.cc/en/research/short-replies/ Thu, 22 Oct 2026 00:00:00 GMT Most of what a caller says is a few words, and a few words are the hardest case for speech recognition. Oasis Speech 2 cuts errors on one-word replies by 49%. Research Speech recognition How we grade a conversation between a voice agent and a simulated caller https://dialogist.cc/en/research/how-we-grade-a-conversation/ https://dialogist.cc/en/research/how-we-grade-a-conversation/ Wed, 21 Oct 2026 00:00:00 GMT Polygon, our test bench, grades a voice agent on whole conversations, with simulated callers, eight weighted criteria, hard gates and behavior probes. Research Evaluation Testing Cadence: the workbench where we fine-tune, evaluate and gate the models behind Oasis Speech https://dialogist.cc/en/research/cadence/ https://dialogist.cc/en/research/cadence/ Tue, 20 Oct 2026 00:00:00 GMT Cadence is where we fine-tune and evaluate the models behind Oasis Speech. People and coding agents use one API, and every model must pass a gate. Research Speech recognition Training What happens to a sentence after the voice model speaks it https://dialogist.cc/en/research/after-the-voice-model/ https://dialogist.cc/en/research/after-the-voice-model/ Mon, 19 Oct 2026 00:00:00 GMT Between the voice model and the caller, the agent trims, resamples, checks and caches every sentence, and re-checks the cache at night. Research Speech synthesis Latency Knowing when the caller has finished https://dialogist.cc/en/research/knowing-when-the-caller-has-finished/ https://dialogist.cc/en/research/knowing-when-the-caller-has-finished/ Wed, 14 Oct 2026 00:00:00 GMT Across nearly 100,000 caller turns, our end-of-turn model closed most turns before the silence timer and gave back about 0.3 s per turn on real calls. Research Turn-taking