最有意思的AI音乐模型，让创作不再差点意思
Ranking
Overall
57
Content
60
Popularity
N/A
No observed public metrics; popularity remains neutral/archived.
Merged summary
TL;DR — 音超 V4.0 是重新训练的新一代 AI 音乐模型，重点提升提示词理解、情感表达、曲风还原、多语言生成和编曲质量，并覆盖大众、专业及企业场景。
- 采用重新设计的架构，而非基于 V3.5 增量升级。
- 支持十大语言，并向普通用户开放纯音乐生成。
- 改善乐器分离、编曲细节，以及对口语化和强情绪提示词的遵循能力。
- 完全使用壁仞国产 GPU 训练，已应用于音超 App、Studio、Hitto 和企业 API。
注： 智东西的摘要实际介绍的是百度 AI4S 与 Famou 决策智能体，和音超 V4.0 并非同一项工作，因此未将其技术内容混入总结。
Sources (2)
最有意思的AI音乐模型，让创作不再差点意思
Public signals
N/A
TL;DR - Chinese AI music company YinChao released its V4.0 model, claiming improved prompt understanding, emotional expression, genre fidelity, and multilingual generation. The release expands AI music creation across consumer, professional, overseas, and API products.
- V4.0 was retrained with a redesigned architecture rather than incrementally updated from V3.5.
- It supports ten major languages and opens instrumental-only generation to general users.
- The model reportedly improves instrument separation, arrangement detail, and adherence to conversational or emotion-heavy prompts.
- YinChao says the model was trained entirely on Biren domestic GPUs; its tools include YinChao App, Studio, Hitto, and an enterprise API.
一个应用，覆盖一万种场景：寻找AI落地“最优解”
Public signals
N/A
TL;DR - Baidu introduced an AI4S “1+10+N” strategy centered on Famou, a self-evolving decision agent that combines LLM reasoning with evolutionary search. It targets complex optimization across scientific research and industrial operations.
- The initiative includes one agent platform, ten interdisciplinary research成果, and a scientist program supporting 100 researchers.
- Famou searches, evaluates, and iterates over candidate solutions under expert-defined objectives, constraints, and metrics.
- Reported applications include neural architecture design, theorem proving, robotics control, port scheduling, chip design, energy management, and risk modeling.
- In pine-disease detection, feature search reduced 107 candidates to about five, raised early-tree detection from 89.1% to 98.0%, and cut false positives from 13.6% to 5.1%.