Kannada-English Adult Audio Lessons

@tratuan

Two things up front so you can rule me out now if I'm wrong for this. The voice would be synthesised, not recorded by a human. I build the audio pipeline; I am not a voice actor. If you want a person in a booth, say so and I'll step aside - no hard feelings. And I don't speak Kannada. I can produce it, but I can't hear a pronunciation error the way you will. So I verify it differently: I run speech recognition back over every generated file and compare what the recogniser reads against the script it was meant to say. Anything that doesn't read back correctly gets flagged and regenerated. That catches wrong pronunciation. It does not tell me whether the intonation sounds natural to a Kannada ear - for that you or a native speaker has to listen. What I am good at is the part your technical notes describe. Consistent volume across every track is not something a human recording 100 files gets right by hand. I measure each file to the same RMS and peak target and hand you the table. 44.1 kHz WAV, noise floor measured rather than assumed, and the pacing set identically on every track so the hear-repeat-absorb rhythm doesn't drift between file 3 and file 80. Scripts come with it. Start with three paid sample tracks. Your ear decides before either of us commits to a series. If the Kannada isn't good enough, you've lost very little - I'd rather find that out at track 3 than at track 80.