Tagged articles
speech processing
10 articles · Page 1 of 1
The article explains why raw audio cannot be fed directly to language models, outlines the two essential compression steps, compares three common tokenization approaches—neural codecs, self‑supervised clustering, and continuous vectors—and warns of typical pitfalls for newcomers.
Self-Supervised Learningaudio tokenizationlarge language models
0 likes · 6 min read

Google researchers discovered a linear relationship between brain activity recorded during natural conversation and the internal embeddings of a speech‑to‑text large language model, revealing that acoustic and lexical representations from the model can accurately predict neural responses in both language comprehension and production.
AI researchGooglebrain imaging
0 likes · 8 min read

Tencent AI Lab deputy Yu Dong was named a 2022 ACM Fellow for his groundbreaking work in speech processing and deep‑learning applications, boasting over 100 patents, multiple best‑paper awards, and technologies now embedded in many of Tencent’s products.
ACM FellowTencentartificial intelligence
0 likes · 5 min read

This paper presents an end‑to‑end approach for extracting relational triples directly from speech signals, bypassing intermediate transcription, and demonstrates its effectiveness on synthesized speech versions of the CoNLL04 and TACRED datasets, highlighting challenges such as length constraints and cross‑modal alignment.
End-to-Endmultimodalnatural language processing
0 likes · 17 min read

This article explains a deep‑learning pipeline that preprocesses audio, extracts FBank features, applies SpecAugment, and uses a CNN‑BI‑LSTM‑Attention model to automatically identify pornographic and ASMR speech for content moderation.
ASMR detectionAudio ClassificationSpecAugment
0 likes · 8 min read

This article walks through a hands‑on experiment using Huawei Cloud RDS and Python to convert short videos into audio, extract voiceprints, and match them against a database, demonstrating how speech‑signal processing enables accurate identification of background songs in TikTok‑style clips.
AIPythoncloud computing
0 likes · 5 min read


58 TechNov 16, 2020 · Artificial Intelligence
This article details the evolution of the voice endpoint detection (VAD) module in 58.com’s voice robot, comparing a dual‑threshold method, Google’s WebRTC VAD, and the deep‑learning based VADNet, and presents experimental results on accuracy, recall, F1 score and online latency.
VADVoice Activity Detectionreal-time communication
0 likes · 14 min read

JD AI Research Institute presented four accepted Interspeech 2020 papers—covering sound event localization, speech dereverberation, speaker verification, and an efficient WaveGlow vocoder—demonstrating significant advances in audio AI despite the conference’s shift to an online format due to COVID‑19.
audio AIneural vocodersound event detection
0 likes · 8 min read


58 TechJul 31, 2020 · Artificial Intelligence
This article presents 58.com’s intelligent voice quality inspection system, detailing its overall architecture, speech separation, speaker role identification, NLP‑based tagging, model choices such as VGG, BERT, ALBERT and SPTM, and real‑world call‑center use cases that improve efficiency and reduce risk.
AINLPcall center
0 likes · 20 min read


58 TechMay 28, 2019 · Artificial Intelligence
This article details the architecture and implementation of the voice call module of an intelligent voice robot, covering SIP signaling establishment, RTP session handling, audio encoding/decoding, sampling, and packetization to enable automated outbound calls and multi‑round voice interactions.
AISIPaudio codec
0 likes · 9 min read
