Verify the reference
Canonical text, IAST, source work, chapter/verse, and provenance remain immutable inputs to the audio workflow.
LISTENING JOURNEY
Canonical text first. Reviewed audio when it is truly ready.
SlokaSandra is building a listening experience around verified scripture rather than attaching generic audio to every page. A track becomes playable only after rights, pronunciation review, provenance, timing metadata, and synchronized playback are complete.
Review the recitation
The Bhagavad Gita 2.47 pilot now has a paid-tier Azure prebuilt female Hindi voice selected for generation testing. It is not a Sanskrit-specific voice and cannot become verified recitation without human Sanskrit pronunciation review, recorded rights metadata, and phrase-level approval.
Synchronize and stream
Reviewed audio is uploaded to private R2, phrase timings bind to canonical IDs, and same-site range-capable delivery enables seek, highlighting, and controlled auto-scroll.
PILOT REFERENCE
Bhagavad Gita 2.47
कर्मण्येवाधिकारस्ते मा फलेषु कदाचन ।
मा कर्मफलहेतुर्भूर्मा ते सङ्गोऽस्त्वकर्मणि ॥
The synchronized player foundation is already attached to the canonical page. The four canonical phrase IDs are packaged for human pronunciation review, and a pilot-generation voice candidate is now recorded. Production audio remains intentionally disabled until a generated or recorded track passes human Sanskrit review and carries commercial-use provenance, private-R2 identity, duration, approved read-along, and phrase timings.
Why some pages are still silentSilence is a quality state, not a missing feature. SlokaSandra keeps a page at “Awaiting verified recitation” rather than publishing uncertain pronunciation, unclear rights, fabricated timings, or an unreviewed synthetic voice.