tushar-dadlani / shruti Goto Github PK

listen to all those updates using Ivona TTS

This is a work-in-progress. In PRE PRE PRE ALPHA stage. A lot of things are expected to be changed.

• shruti is a simple server that keeps listening for new TTS requests
• various other programs do the labour. Fetch updates, scrape, consume api or whatever. Then they push their notifications to shruti
• shruti will get the TTS part done using Ivona and push it to its queue.
• After specified interval, shruti will make sure to read out those notifications
• one after another

• I am bored of reading everything myself. I would love if someone does the job for me
• Something that will check
  • twitter. My timeline and few accounts I like to keep up with
  • HN, few reddit channels
  • some headlines from bbc
  • some important server statuses
  • build server status updates if important
  • New email upon arrival. From and Subject
  • Share market updates. Few important share values I care about
  • any any other sites I can think of
• ... and read out loud every 15-30 minutes.
• Hence (:

• The shruti is a Sanskrit term, which in the context of Indian music, is considered the smallest interval of pitch that the human ear can detect
• Shruti - wiki
• besides, its a girl's name, suits the voice and sounds catchy enough

• Use any language, do whatever to get the text
• Send an HTTP request to shruti server at http://host:port/push/
  • parameters required:
    • text : string : text to be spoken
    • voice : string : ivona voice [not implemented yet]
    • provider : string : provider name for reference
    • keepfile : boolean : yes/no to keep file [not implemented yet]

• sign up for beta account on ivona.com
• generate credentials
• install ffmpeg. We need ffplay binary
• get the shruti server binary, make sure to keep it up and running
• make sure to export environment variables SHRUTI_ACCESSKEY and SHRUTI_SECRETKEY
• run your providers however you want
• You could use supervisor or upstart on Ubuntu.

• Build an open source SaaS that anyone can deploy for themselves, use it personally

• Do not go after building J.A.R.V.I.S. There is already a project for that jasper

• TODO:

  • Convert this hack into a nice future proof API, make it configurable easily
  • Support multiple audio player backends to make it usable on Windows and Mac
  • Add providers
  • Add actions, Mute, Unmute, Force speak
  • Build configuration, status pages
  • Make use of different voices for different providers
  • Add support for noticing special tags probably. Add extra text to such texts to make it more natural

Contributions, thoughts are welcome :)

License MIT