目前，最新语音同传大模型已经在讯飞双屏翻译机2.0、讯飞AI录音笔、讯飞同传等产品中上线，并且作为世界人工智能大会(WAIC)翻译合作伙伴，还为大 会提供AI翻译服务，助力实现无障碍沟通。 才灏指出，讯飞深耕多年的智能语音创新算法和大模型形成了很好的互补。"这些算法在语音信号的前端处理上特别有优势，比如降噪、回声消除、方言口 音适配等，跟大模型结合之后，能让前端的语音信号处理更精准、更高效，为大模型后续的语言理解、语义分析打下更扎实的基础，从而让整个智能交互体 验更流畅、更可靠；大模型也为智能语音技术带来了更好的效果和更多的可能"。 基于这些创新技术在智能硬件的应用，科大讯飞做到了软硬件一体化，比如新一代翻译机双屏翻译机2.0通过硬件性能的持续升级，显著提升了抗干扰能力 与耐用性。这种软硬件深度融合的专业设计，能够屏蔽1米外的噪音，在展会洽谈、工厂巡场的嘈杂环境中也能使用，更适用用户的实际使用场景，使其与 7月26-28日，2025世界人工智能大会暨人工智能全球治理高级别会议("WAIC2025")在上海举行，来自全球30余个国家和地区的1200余位嘉宾，超800家企业 重磅参展，3000余项前沿展品集中亮相 ...

Group 1: Overall Business Performance - The company's total revenue for 2024 reached 233.43 billion CNY, an increase of 18.79% compared to 196.50 billion CNY in 2023 [6] - Gross profit for 2024 was 99.52 billion CNY, reflecting an 18.70% growth from 83.84 billion CNY in 2023 [6] - Net profit attributable to shareholders decreased by 14.76% to 5.60 billion CNY from 6.57 billion CNY in the previous year [6] - The net profit after deducting non-recurring gains and losses increased by 59.32% to 1.88 billion CNY from 1.18 billion CNY [6] - Operating cash flow reached a record high of 24.95 billion CNY, a significant increase of 613% from 3.50 billion CNY in 2023 [10][11] Group 2: Quarterly Performance - In Q4 2024, revenue was 70.37 billion CNY, a 21% increase from Q4 2023 [8] - Cash flow for Q4 2024 was 33.16 billion CNY, up 118% from Q4 2023 [8] - Net profit for Q4 2024 was 9.04 billion CNY, a 62% increase compared to 5.58 billion CNY in Q4 2023 [8] Group 3: Research and Development Investment - In 2024, the company invested 7.44 billion CNY in large model research and application, maintaining a R&D expenditure ratio of around 20% of revenue [17] - R&D investment for 2024 was 45.80 billion CNY, slightly increasing from 38.36 billion CNY in 2023, with the R&D expenditure ratio at 19.62% [18] Group 4: Business Structure and Strategy - The company has optimized its business structure, with the proportion of revenue from core business areas increasing by 5% [15] - The number of operational bases increased from 54 in 2022 to 60 in 2023, and further to 46 in 2024, indicating a significant expansion [14] - The company emphasizes a self-controlled core technology strategy, which is crucial for maintaining competitive advantages in the AI sector [16][61] Group 5: AI Model Performance - The Spark-X1 model has shown significant improvements in various tasks, outperforming competitors like OpenAI and DeepSeek in text generation and language understanding [24] - In the medical field, the Spark medical model achieved a diagnostic accuracy rate of 94.0%, surpassing GPT-4o and DeepSeek R1 [30] - The legal model demonstrated a comprehensive accuracy rate of 82.89% across nine legal tasks, with key tasks exceeding 90% accuracy [34] Group 6: Market Position and Future Outlook - The company is positioned as a leader in various sectors, including education, healthcare, and enterprise applications, with a focus on personalized AI solutions [70] - The competitive landscape is intensifying, with a need for self-sufficiency in AI technologies due to increasing geopolitical tensions [61][66] - The company aims to leverage its advancements in AI to enhance productivity and foster innovation in various industries [74]

Performance Overview - In Q1 2025, iFlytek's revenue grew nearly 30%, with gross profit increasing by 35% and business opportunities rising by almost 50% [4] - The company achieved a historical high in cash flow for 2024, with a net cash flow from operating activities of CNY 2.495 billion [21] - The overall gross margin remains stable, with a projected revenue growth of approximately 27% and gross profit growth of about 35% in Q1 2025 [22] AI Model Development - iFlytek's Starfire X1 model, based on fully domestic computing power, is recognized as the best deep reasoning model among 70B parameter models, achieving performance comparable to international top models [6][17] - The company emphasizes the importance of high precision and reliability in key areas, focusing on closed-source systems while also embracing open-source models [5] - The Starfire model's training efficiency has improved significantly, with training efficiency rising from 55% to 85% compared to A100 [6][20] Market Demand and Strategy - The demand for domestically controlled AI solutions is increasing, especially in sectors like government, education, and healthcare, where iFlytek is well-positioned [8][24] - The company is actively expanding its presence in the education sector, with a focus on AI-driven personalized learning solutions [10][12] - iFlytek's hardware products, such as the AI learning machine and smart office notebook, have shown significant revenue growth, with the smart office notebook's revenue increasing by 50% in Q1 [14] Geopolitical Context - The ongoing US-China trade tensions are not expected to significantly impact iFlytek's operations, as the company has developed a self-sufficient supply chain and primarily serves the domestic market [23] - The company anticipates that the demand for autonomous AI solutions will grow in response to geopolitical challenges, positioning itself to capitalize on these opportunities [24] Future Outlook - iFlytek plans to enhance its international business through the development of multilingual models, supporting various languages to meet global market needs [18] - The company aims to leverage its technological advantages in the healthcare sector, focusing on AI-assisted medical solutions and expanding its partnerships with top hospitals [15][16]