feat(server): refactor record schema (#14729)

#### PR Dependency Tree


* **PR #14729** 👈

This tree was auto-generated by
[Charcoal](https://github.com/danerwilliams/charcoal)

<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit

* **New Features**
* Transcriptions now produce structured meeting summaries (strict JSON)
and a normalized, speaker‑tagged, non‑overlapping transcript with legacy
projection support.

* **API**
* Submission accepts richer transcription input; results return
source‑audio metadata, slice manifest, quality indicators, normalized
segments/transcript, and structured summary JSON.

* **Frontend**
* Recording flow stores transcription metadata and uploads preprocessed
audio slices with slice/quality info; UI-side result normalization
applied.

* **Tests**
* Expanded unit, contract, and e2e coverage for normalization, payload
parsing, persistence/retry, and end‑to‑end transcription flows.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->

packages/frontend/core/src/blocksuite/ai/blocks/transcription-block/model.ts

@@ -1,3 +1,7 @@
+import type {
+  TranscriptionQualityInput,
+  TranscriptionSourceAudioInput,
+} from '@affine/graphql';
 import {
   BlockModel,
   BlockSchemaExtension,
@@ -25,7 +29,10 @@ export const TranscriptionBlockSchema = defineBlockSchema({
 });
 
 export type TranscriptionBlockProps = {
-  transcription: Record<string, any>;
+  transcription: {
+    sourceAudio?: TranscriptionSourceAudioInput;
+    quality?: TranscriptionQualityInput;
+  };
   jobId?: string;
   createdBy?: string;
 };