Course Narration Studio

Problem

Course voiceover often treats a whole course as one long audio file, causing inconsistent terminology, slide labels being read aloud, and costly full-lesson re-recording. @beatra-ai/course-narration-studio targets courses, training, or internal materials by turning finalized scripts into lesson-level narration packages rather than one long audiobook.

How It Works

• Build a lesson ledger: capture course name, order, one teaching point per section, and a pronunciation table; freeze the pilot lesson's speakable text.
• Voice selection and synthesis: with a narrator sample, confirm authorization before beatra.voices.clone; without a sample, use beatra.voices.list to choose a catalog voice and synthesize with beatra.speech.synthesize.
• Pilot first: choose the hardest lesson for review, then revise only the misread or unclear blocks instead of re-recording the whole lesson.
• Expand in batches: after pilot approval, confirm the remaining estimate, submit blocks individually, poll beatra.tasks.get, and deliver labeled audio with duration, usage, and billing fields.

Boundaries

Best for multi-lesson courses that need one consistent teacher-like voice. Do not use it for literary works, long-form narration, clone series, talking-head course videos, or short voiceovers. Strip page titles, bracket labels, and image captions from input; never guess pronunciation for repeated names, terms, or abbreviations. Cloning and each synthesis block are billable, and insufficient balance should be resolved before retrying with the frozen ID.