-
Voice query
-
Text query
Inspiration
We decided to use different APIs to learn how they work. Telegram and SkyScanner were the perfect combination for our needs because they both have a wrapper for Python and we thought that it would be useful to ask for a flight directly to Telegram.
What it does
It searches for the next flight to the cities indicated to the Telegram bot via text or via a voice note.
How we built it
We used python as a programing lenguage, and a bunch of libraries and dependances like the Telegram Python wrapper for the Telegram bot comunication, the SpeechRecognition library for the speech function, the AudioSegment wrapper for transform audio files to be compatible with the differents libraries and ffmpeg framework as a dependence of AudioSegment. Also we used Hidora to host the application.
Challenges we ran into
One of the firsts challenges has been to implement the SpeechRecognition funcionality because the Telegram bot uses a type of audio file that is incompatible for the SpeechRecognition library and we needed to convert the files with the ffmpeg framework and we had difficulties to deploy a docker with this framework installed in it. Also, another challenge was to read and understant the documentation of the API's and the functionalities of them that weren't in the documentation.
Accomplishments that we're proud of
Being able to build the entire bot and all it's dependencies in a docker for an seamless deployment
What we learned
Some members of the team hadn't worked with APIs and dockers before, so we learned about them. It was quite a challenge at first but finally we could understand them enough to work with as a team.
What's next for SkyTelegram
Optimize the usage of the Skyscanner API so we can give a faster response, recognize a date and some preferencies from the user in order to offer a suitable flight, not only answer with the cheapest flight and bring translations, so the user can choose between many languages.
Log in or sign up for Devpost to join the conversation.