Search inside audio files

Search for words inside audio files or Telegram voicemails. Powered by Deepgram. Requires API keys from Deepgram and optionally Telegram. Submission for Deepgram+DEV hackathon, 2022.

You might want to read the submission post.

Get the API keys

Deepgram (required): Create an account in deepgram.com and get an API key.
Telegram (optional): Create an account in Telegram and follow the steps here: Obtaining api_id

Store them in files named deepgramApiKey, telegramApiId and telegramApiHash in the root folder or pass them directly in the CLI using the --deepgram-api-key, --telegram-api-id and --telegram-api-hash arguments.

Features

Tune the voice recognition process with the Deepgram query parameters for transcriptions pre-recorded audio with -P|--param KEY=VALUE arguments.
Search directly in local files passing them as arguments after the search term.
Automatically download audios from chats in Telegram with one or more -T|--telegram-chat CHAT_ID arguments.
Downloads and results are cached to reduce redundant traffic, but you can force it using the -F|--force flag or directly removing the _cache and _audio folders.
Search for partial matches or for whole words using the -W|--whole-word flag.
Include a bit of the context in which the word was said for each hit.
All log information outputs through stderr and the search output through stdout (or a file, with the -o|--output-file FILE argument). This makes it easy to redirect and pipe different information safely.

Output in the following JSON format:

[                        // list of audio files with matches
  {
    "source_file":str,   // path to the audio file
    "duration":num,      // duration of the file
    "hits": [            // list of hits in the file
      {
        "position":num,  // position of the word in the transcript
        "start":num,     // start time in the audio
        "end":num,       // end time in the audio
        "context":str,   // text in the transcript surrounding the match
      },
      ...
    ]
  },
  ...
]

Quick start

Install all dependencies (telethon too):
```
pip install -r requirements.txt
```
Search for a term in local files:
```
python main.py TERM FILES...
```
Search for a term in audios from chats in Telegram:
```
python main.py TERM -T CHAT1 -T CHAT2 ...
```
Print all available options:
```
python main.py -h
```

Contributing

Contributors

Empty section, for now ;)

Ideas for the future

Add more remote audio sources, apart from Telegram chats (maybe Discord?).
Make the search process more flexible using an edit-distance based match, instead of only exact matches.
Allowing more complex queries: multiple words, regular expresions, etc.
If you can think of another one, feel free to make a PR!

License

This code is licensed under the MIT license, a copy of which is available in the repository.