| I&T Solution |
AI smart voice-to-text solution
(REF: S-1305) |
| Trial Project |
|
| Solution Feature |
- The voice signal comes in, and then goes through noise reduction, endpoint detection, and feature extraction. This is the front-end processing module.
- After front-end processing, after obtaining the features, you can enter the recognition process, one is the acoustic model, and the other is the language model;
- Acoustic model language model, recognize and decode in the decoder to get the corresponding recognition result
- The recognition result is subjected to some text processing related operations, and finally the recognition result is obtained
- During offline training, the extracted speech features will be
|
| Trial Application and Expected Outcome |
- Realize millisecond-level recognition for audio streams, and return text streams with timestamps for secondary development;
- Supports instant transcription in 8 languages, 12 dialects and 2 national languages
- Perform semantic understanding on the context, and intelligently correct the intermediate results to ensure high recognition accuracy;
- Returns formatted text for digits, dates, times, etc., intelligently segmenting sentences and matching punctuation according to the context of the dialogue.
- Returns formatted text for digits, dates, times, etc., intelligently segmenting sentences and matching punctuation according to the context of the dialogue.
|
| Additional Solution Information |
AI智能语音转文字解决方案-繁（使用第三方实时语音转写API接口和模型）.pdf
|
| Info on I&T Solution Provider |
| Solution Provider | : | Hong Kong Zi Wai Technology Limited | | Address | : | Room 702, 7/F, Fu Fai Commercial Centre, 27 Hillier Street, Sheung Wan, Hong Kong | | Contact Person | : | 張永強 |
| Position | : | General manager | | Tel | : | 35212970 | | Email | : |
hkziwai@gmail.com | | Webpage | : | www.hkziwai.com |
|
