Core Speech Kit integrates basic AI voice capabilities, including text-to-speech (supported by the TextToSpeech module) and speech recognition (supported by the SpeechRecognizer module). It facilitates user interaction with devices by enabling real-time conversion between spoken language and written text.
Text-to-speech: Synthesizes a text containing a maximum of 10,000 characters into speech and plays the speech. For details about the implementation effect, please refer to TextReader.
Speech recognition: Converts a piece of audio information (no more than 60s in short speech mode and no more than 8 hours in long speech mode) into text. You can convert a PCM audio file or real-time audio into text. For details about the implementation effect, please refer to AICaption.
| AI Capability | Constraint |
|---|---|
| Text-to-speech | - Supported languages: Chinese and English (Simplified Chinese, Traditional Chinese, and English within a Chinese context). - Supported timbre types: Ling Xiaoshan female voice, Laura female voice for English (US), and Ling Feizhe male voice. - Text length: no more than 10,000 characters |
| Speech recognition | - Supported language: Mandarin Chinese - Supported model type: offline - Speech duration: no more than 60s in short speech mode and no more than 8 hours in long speech mode |
This kit supports emulators since version 6.0.0(20).
There are some general differences between the Emulator and real devices. For details, please refer to Differences Between the Emulator and the Real Device.