How Outbound Call Robots Work: Challenges and Optimizations in Voice Dialogue Systems

This article explains the architecture of outbound call robots, classifies dialogue system types, details pipeline and end‑to‑end task‑oriented designs, highlights technical challenges such as dialects and transcription errors, and presents optimization techniques like ASR correction and script improvement.

AI OptimizationASR correctionNLU

0 likes · 12 min read

How Outbound Call Robots Work: Challenges and Optimizations in Voice Dialogue Systems

DataFunTalk

Feb 18, 2021 · Artificial Intelligence

Didi Voice Interaction: ASR Error Correction, Intent Classification, and NER Techniques

This article presents Didi's voice interaction platform, detailing the natural language understanding pipeline, ASR error correction methods, intent classification strategies, and named entity recognition models, while discussing practical deployments, performance gains, and future research directions.

ASR correctionSpeech interactionintent classification

0 likes · 18 min read

Didi Voice Interaction: ASR Error Correction, Intent Classification, and NER Techniques

Didi Tech

Nov 18, 2020 · Artificial Intelligence

Didi Speech Interaction: ASR Error Correction, Intent Classification, and NER Techniques

Didi’s voice‑interaction platform combines a three‑stage ASR error‑correction pipeline, optimized intent‑classification models (both end‑to‑end and retrieval‑based), and advanced Chinese NER using Bi‑GRU‑CRF and BERT‑CRF, boosting transcription accuracy and overall dialogue success while supporting scalable deployment and future enhancements such as lattice inputs and richer acoustic signals.

ASR correctionSpeech interactionintent classification

0 likes · 21 min read

Didi Speech Interaction: ASR Error Correction, Intent Classification, and NER Techniques