DeepSeek App Unveils Beta Testing of Voice Function and Four Reading Tones

Multiple users reported that a voice function entry appeared in the top right corner for some DeepSeek App users; additional reports indicate that a "Reading Tones" option was added to some users' settings page. This suggests that a voice-related interface in DeepSeek App has begun to appear for some users, but currently the evidence is based on user feedback and reports, not an official product announcement.

According to the known timeline, DeepSeek previously announced that DeepSeek V4.1 Flash officially launched on September 10, claiming the model has native multi-modal visual understanding capabilities. DeepSeek also stated that V4.1 Flash surpassed various flagship models, including DeepSeek V4 Pro, in their benchmark tests. The timing of this model release closely coincides with changes in the app interface, but the two should not be directly considered components of the same feature.

The information directly related to reading focuses on the four tones in the settings page: Shell, White Wave, Starfish, and Dark Tide. Reports describe them as versatile and lively, bright and firm, playful and sweet, and deep and robust, respectively. The different tones indicate that the app is at least attempting to make the voice presentation of text content a selectable experience rather than just offering a single default voice.

When assessing this situation, it's important to separate the sources of information: the voice entry and tone options come from user feedback and reports, while the release date, visual understanding capabilities, and benchmark performance of V4.1 Flash are from DeepSeek's statements. These two types of information together form the current observational material, but cannot infer that the voice function is tied to a specific model version, nor can one equate the user-side interface directly with an official feature available to all users.

Therefore, the most accurate conclusion from this information is that DeepSeek App has been observed to feature a voice entry and related interface for four reading tones. For readers interested in this function, the focus should not be on prematurely viewing it as a complete product release, but rather on distinguishing the interface changes captured by user feedback from the model capabilities publicly stated by DeepSeek.