Millions of meetings are held and decisions made verbally on a daily basis. We generate a summary to help people remember what they spoke about, and those who missed the meet /call understand it.
Algorithms Used: Glove, Encoder-Decoder, Attention Mechanism API’s Used : Google Speech to text Programming Language : Python
Step 1: Speech – The input is the Speech from the user which is converted to text and is stored in the text File. Audio File – the audio file is sent as an input where it will get converted in to text with the help of google APIs.
Step 2: In this Step, We Convert the input audio file or speech to the Google API. This API converts all the Speech to text and store it in a text file. We use libraries like PyAudio, Pyttsx to use the microphone service and other features related to Speech to text conversion.
Step 3: Clean the data of any special characters and lengthy spaces to make clean sentences. Calculate frequency of every word and finds words with high number of occurrences. Sentences with words that have high number of occurrences will get the highest priority. These sentences will lead to a short summary as an output.
Advantages: Easy to follow up the conversation that has been made. Tracking the progress of projects underway. Efficient and reliable output after continuous training.
Risks: Output is a text file and hence needs to be protected. Accuracy cannot be guaranteed of noisy data.
Future Improvements: Multi-lingual interface. Video input