spike: existing offline voice app integration

[SebastianBoehler]

Issue Spec: Existing Offline Voice App Integration Spike

Summary

Check whether an existing offline voice app, speech-to-text tool, or local voice framework can be reused instead of building the full voice stack ourselves.

Goal

Avoid unnecessary voice infrastructure work if a reliable local/offline integration already exists.

Scope

• Identify 2-3 realistic offline voice or speech recognition options.
• Check platform support, license, integration effort, model size, and maintenance risk.
• Test or document whether recognized commands can be passed into StudyOS locally.
• Compare against native assistant shortcuts and app-owned push-to-talk/hotword input.

Out Of Scope

• Committing to a production dependency before comparison.
• Training custom speech models.
• Using hosted speech APIs as the default path for sensitive student data.

Acceptance Criteria

• At least two candidate tools/apps/frameworks are compared.
• The comparison includes offline support, platform coverage, privacy, license, integration effort, and expected UX.
• One minimal integration path is described or prototyped.
• The output includes a recommendation: reuse, wrap, defer, or avoid.

Gates To Pass

• Privacy gate: offline mode is verified or clearly marked as not available.
• License gate: the candidate license is compatible with the project.
• Platform gate: mobile target support is checked.
• Integration gate: there is a realistic way to pass recognized commands into StudyOS.
• Maintenance gate: model size, updates, and app packaging cost are considered.
• Decision gate: the team can decide whether this saves enough work to justify dependency risk.

Suggested Branch

spike/offline-voice-integration