EIGEN JOURNAL

Notes from the edge of voice AI.

Notes from the edge of voice AI.

Notes from the edge of voice AI.

Engineering perspectives, releases, and practical ideas for building clearer, more reliable conversational systems.

Engineering perspectives, releases, and practical ideas for building clearer, more reliable conversational systems.

Research

Eigen on Noisy Hindi Audio: Kathbath WER with Deepgram Nova 3 and Soniox v5

Hindi Kathbath WER with Deepgram Nova 3 and Soniox v5 before and after Eigen processing.

Read article

Research

Cartesia ink-2 with Eigen: WER Results Across Three Datasets

Cartesia ink-2 WER before and after Eigen across three reported datasets, with separate latency testing recommended.

Read article

Research

Soniox v5 with Eigen: 44.3% Lower WER Than Baseline on Real World Audio

Soniox v5 WER before and after Eigen, including a 44.3% relative reduction on the Real World Audio set.

Read article

Research

AssemblyAI Universal 3 Pro with Eigen: WER Results Across Three Datasets

AssemblyAI Universal 3 Pro WER before and after Eigen across three reported datasets.

Read article

Research

Deepgram Nova 3 with Eigen: WER Results Across Four Datasets

Deepgram Nova 3 WER before and after Eigen across four datasets, including noisy Hindi and partner-call audio.

Read article

Research

Eigen, Krisp, and ai-coustics: WER on the Real World Audio Set

A paired WER comparison of Eigen, Krisp VIVA 2.0, ai-coustics Quail VF 2.2L, and unenhanced audio on one partner-call set.

Read article

Research

Eigen Benchmark: WER Results for Four ASRs Across Selected Datasets

WER results for Eigen across four streaming ASRs and selected clean, structured, noisy Hindi, and partner-call datasets.

Read article