HTWCore is based on a .NET Core of winform client program can be used to handle a variety of meetings, records, seminars, lectures and so on audio and video sources, the use of speech recognition, video content extraction techniques organized into a word document. Projects using the following technologies:
- Based on .NET Core, and hence is a cross-platform program that can be used for various versions of windows platform, the future will fit linux, mac platforms. The packaged into a single file, from the installation, the performance is relatively high.
- Use audio segmentation, volume processing, audio and video decoding, transcoding, audio and video noise reduction, video audio and video frame extraction processing.
- Use of the current leading voice recognition technology can identify audio conferencing, recording, lectures and so on scene.
- Using identification techniques to distinguish the characters recorded voiceprint, sorting out the natural, readable document.
- Speech recognition process is recorded in the time code of each word can be used to generate SRT file, add a subtitle file.
- Using text analytics, natural language processing technology, the type of word processing, natural add punctuation.
- Extraction video embedded subtitles first separated into a valid text images containing video frame, and then based on the similarity with the SSIM (structural similarity is a measure of similarity of the two images) algorithm for comparing two images, removed similar high pictures and improve performance.
- The video frame OCR, can handle a variety of natural scenes with text images, can be Chinese, English, Japanese, Korean, etc., can also be a bilingual text, bilingual branch extraction results show.
- The contents of the text using text similarity algorithm, effectively de-emphasis.
- The results will be exported as text processing software commonly used word, word wrap, indents, document content easy to read.
HTWCore effect is as follows:
The audio and video file into a text box.
- It can be directly identified or may be audio and video processing.
- Click the "Word" button, progress indication.
- The results generated by the word processing document in the same directory of audio and video, one hour of audio takes less than 10 minutes the result.
Good voice quality recognition accuracy rate above 95%.
- Embedded subtitles, the first shots framed range.
Adjusting the left frame appears elliptical text subtitle right slider adjustment range.
其余功能可直接使用，不一一说明了。
HTWCore下载：
链接: https://pan.baidu.com/s/12A7ot2kZkYIoekPi5U9tMw 提取码: es72