[pylrn]

Summary

Expands the Speech Understanding configuration types in the Node SDK to achieve full parity with AssemblyAI's API and the Python SDK:

• SpeakerIdentificationRequest.speakers: Accepts an array of SpeakerDescription objects (name, role, description, etc.) for rich speaker identification, providing the typed alternative to known_values when identifying speakers by name or role with context.
• TranslationRequest.force_translation: Adds the force_translation?: boolean parameter to allow translating transcripts even when the detected source language matches a target language.
• CustomFormattingRequest.format_utterances: Adds the format_utterances?: boolean parameter to apply custom formatting at the utterance level.
• Exports SpeakerDescription: Exported from package root via ./types.

Verification

• npx tsc --noEmit: Clean, 0 type errors.
• npx rollup -c: Builds all targets (ESM, CJS, UMD, browser, Node, Bun, Deno, workerd) cleanly.
• Unit tests added in tests/unit/speech-understanding-effort.test.ts verifying speakers (name and role modes), force_translation, and format_utterances: 10/10 tests passing.