The content of this page has been automatically translated by AI. If you encounter any problems while reading, you can view the corresponding content in Chinese.
This article will introduce how to experience the recognition feature through the ASR console. You can experience the recording file recognition feature by directly uploading a file or URL link, or experience the real-time speech recognition feature by scanning a code.
Video Explanation
Trying Out the Recording File Recognition Feature
File source: Supports uploading local files and URL links. You need to follow the Recording file recognition requirements in the product details. The uploaded voice file size should not exceed 1 GB and the duration should not exceed 5 hours.
Audio category: Supports call and non-call. The recommended bit depth for both categories is 16-bit. The audio category must match the uploaded audio to get the correct recognition result. If you do not know the audio properties of the recording file, you can check it in common audio software (e.g., Adobe Audition) or use the open-source command line tool FFmpeg.
Call: Audio generated from mobile phone or fixed-line phone calls, generally with a default sampling rate of 8000 Hz.
Non-call: Audio not generated from mobile phone or fixed-line phone calls, with a recommended sampling rate of 16000 Hz.
Recognition type: Supports general ASR and large model ASR.
General ASR: Tencent Cloud General ASR Engine.
Large Model ASR: Tencent's newly launched ASR large model greatly improves identification accuracy across industry datasets.
For supported language types, please visit the Console.
Engine model: You can choose based on the language and industry of your actual audio. If there is no engine model for the corresponding industry, it is recommended to use the general model for the corresponding language for recognition.
Result format: Supports formats with and without timestamps.
With timestamps: The recognition result includes the start and end time of the corresponding voice segment.
Without timestamps: The recognition result contains only text.
Recording file: select file/file address.
When "File Source" is set to Local File, click Select File to upload a local file.
When "File Source" is set to URL Link, you need to fill in the URL address of the voice.
After uploading the file, click Start Recognition. Once the recognition is complete, click Download Result to view the content of the speech recognition.
Click here to go to the Recognition Records page, where you can view information such as audio name, duration, type, engine model, and status.
Trying Out the Real-Time ASR Feature
1. Scan the QR code with your phone to experience the real-time speech file recognition feature.
2. Select "ASR" to enter the feature experience.
3. Choose the engine model you want to experience.
4. Hold the button to speak. Please start speaking only after fully pressing the button and release it after finishing speaking.
5. You can get the recognition result in real time.
