Design and Optimization of a Kaldi‑Based Speech Recognition Backend at 58.com

This article details the evolution from the initial Kaldi‑based speech recognition architecture (version 1.0) to a re‑engineered version 2.0, describing business background, service components, identified shortcomings, and a series of performance, concurrency, GPU, I/O, GC, and dispatch optimizations that dramatically improve resource utilization, latency, and reliability for large‑scale voice processing at 58.com.

AIBackend ArchitectureGPU

0 likes · 15 min read

Design and Optimization of a Kaldi‑Based Speech Recognition Backend at 58.com

58 Tech

Jul 21, 2021 · Artificial Intelligence

Streaming Speech Recognition Engine: Architecture, Workflow, and Optimizations at 58.com

The article details the design, components, real‑time processing flow, and performance optimizations of 58.com’s streaming speech recognition engine, covering its SDK access layer, logical services, data storage, Kaldi‑based decoding, and the practical impact on voice‑driven applications.

AIKaldiSpeech Recognition

0 likes · 12 min read

Streaming Speech Recognition Engine: Architecture, Workflow, and Optimizations at 58.com

58 Tech

Feb 22, 2021 · Artificial Intelligence

Building a Self‑Developed Speech Recognition Engine at 58.com: From Team Formation to Production Deployment

This article details how a three‑person team at 58.com built a self‑developed speech recognition engine in less than a year, covering background, team formation, data annotation, model selection, engineering architecture, performance optimizations, deployment results, and future directions.

ASRKaldiSpeech Recognition

0 likes · 25 min read

Building a Self‑Developed Speech Recognition Engine at 58.com: From Team Formation to Production Deployment

58 Tech

Nov 27, 2020 · Artificial Intelligence

An Overview of Kaldi Chain Model Speech Recognition and Its Relationship with HMM‑DNN and Discriminative Training

This article explains the Kaldi chain model speech‑recognition system, covering HMM‑DNN fundamentals, discriminative (MMI) training, the special single‑state HMM topology, TDNN architecture, training pipelines, and experimental results that demonstrate its performance advantages over traditional GMM‑based approaches.

HMM-DNNKaldiSpeech Recognition

0 likes · 19 min read

An Overview of Kaldi Chain Model Speech Recognition and Its Relationship with HMM‑DNN and Discriminative Training

58 Tech

Aug 19, 2020 · Artificial Intelligence

Speech Recognition in 58.com: Application Scenarios, Data Collection, Kaldi Chain Model Practice, and End‑to‑End Exploration

This article presents a comprehensive overview of how 58.com leverages large‑scale voice data from call‑center, private phone, and micro‑chat platforms, detailing data collection, annotation, Kaldi‑based chain model training, lattice‑free techniques, and end‑to‑end Transformer‑CTC models to improve Chinese speech recognition performance.

ASRChineseEnd-to-End

0 likes · 16 min read

Speech Recognition in 58.com: Application Scenarios, Data Collection, Kaldi Chain Model Practice, and End‑to‑End Exploration

Liulishuo Tech Team

Oct 28, 2016 · Artificial Intelligence

Open‑sourcing kaldi‑ctc: Fast GPU‑Accelerated CTC End‑to‑End Speech Recognition

The article announces the open‑source release of kaldi‑ctc, a GPU‑accelerated CTC‑based end‑to‑end speech recognition toolkit built on Kaldi, warp‑ctc and cuDNN, highlighting its 5‑6× training speedup, real‑time decoding factor of 0.02, and performance comparisons on the LibriSpeech corpus.

ASRCTCGPU

0 likes · 4 min read

Open‑sourcing kaldi‑ctc: Fast GPU‑Accelerated CTC End‑to‑End Speech Recognition