一种语音合成方法、装置、电子设备及存储介质
By filtering similar timbre features of the target speaker in the pre-trained speech synthesis model and performing fine-tuning, the problem of time-consuming and labor-intensive personalized speech synthesis in the existing technology is solved, realizing efficient personalized speech synthesis, which is suitable for single and multi-person scenarios.
CN116312467BActive Publication Date: 2026-07-17DINGFU NEW POWER (BEIJING) INTELLIGENT TECH CO LTD
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- DINGFU NEW POWER (BEIJING) INTELLIGENT TECH CO LTD
- Filing Date
- 2022-11-29
- Publication Date
- 2026-07-17
AI Technical Summary
Technical Problem
Existing speech synthesis algorithms require remodeling when faced with new personalized speech, which is time-consuming and labor-intensive, reducing the efficiency of speech synthesis tasks.
Method used
By acquiring training data of the target speaker, extracting its timbre features, and selecting similar speaker timbre features in the pre-trained speech synthesis model, the pre-trained model is fine-tuned using the fine-tune method to form a fine-tune speech synthesis model, thereby achieving personalized speech synthesis.
Benefits of technology
It improves the training efficiency and overall efficiency of personalized speech synthesis models, supports single-person and multi-person speech synthesis, is highly adaptable, and can quickly adapt to more scenarios.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN116312467B_ABST
Abstract
本申请提供一种语音合成方法、装置、电子设备及存储介质，其中语音合成方法包括：获取目标说话人训练数据，并提取目标说话人音色特征；在预训练数据中筛选与目标说话人的相似说话人音色特征；将训练好的预训练语音合成模型的模型参数加载至finetune语音合成模型；采用相似说话人音色特征与目标说话人音色特征共同训练finetune语音合成模型；将待合成文本输入训练好的finetune语音合成模型进行语音合成任务。通过预选构建的预训练模型，通过finetune的方式对预训练模型进行微调，以满足语音合成任务的及时性需求，极大提升了个性化语音合成模型的训练效率，进而提升了个性化语音合成任务的整体效率。
Need to check novelty before this filing date? Find Prior Art