feat: 将视频转写重构为纯文字异步语音识别

2026-10-03 21:58:06 +08:00
parent 040273586b
commit 6f0e38054b
34 changed files with 2283 additions and 1771 deletions
@@ -24,29 +24,30 @@ model Author {
 }

 model Video {
-  aweme_id      String           @id
-  desc          String
-  preview_title String?
-  duration_ms   Int
-  created_at    DateTime
-  share_url     String
-  digg_count    BigInt           @default(0)
-  comment_count BigInt           @default(0)
-  share_count   BigInt           @default(0)
-  collect_count BigInt           @default(0)
-  authorId      String
-  createdAt     DateTime         @default(now())
-  updatedAt     DateTime         @updatedAt
-  tags          String[]
-  video_url     String
-  height        Int?
-  width         Int?
-  cover_url     String?
-  fps           Int?
-  raw_json      Json?
-  comments      Comment[]
-  author        Author           @relation(fields: [authorId], references: [sec_uid], onDelete: Cascade)
-  transcript    VideoTranscript?
+  aweme_id          String             @id
+  desc              String
+  preview_title     String?
+  duration_ms       Int
+  created_at        DateTime
+  share_url         String
+  digg_count        BigInt             @default(0)
+  comment_count     BigInt             @default(0)
+  share_count       BigInt             @default(0)
+  collect_count     BigInt             @default(0)
+  authorId          String
+  createdAt         DateTime           @default(now())
+  updatedAt         DateTime           @updatedAt
+  tags              String[]
+  video_url         String
+  height            Int?
+  width             Int?
+  cover_url         String?
+  fps               Int?
+  raw_json          Json?
+  comments          Comment[]
+  author            Author             @relation(fields: [authorId], references: [sec_uid], onDelete: Cascade)
+  transcript        VideoTranscript?
+  transcriptionTask TranscriptionTask?

  @@index([authorId])
  @@index([created_at])
@@ -138,18 +139,32 @@ model ImageFile {
 }

 model VideoTranscript {
-  id                 String                   @id @default(cuid())
-  videoId            String                   @unique
-  speech_detected    Boolean
-  language           String?
-  audio_type         String?
-  transcript         String[]
-  non_speech_summary String?
-  createdAt          DateTime                 @default(now())
-  updatedAt          DateTime                 @updatedAt
-  transcript_zh_tsv  Unsupported("tsvector")?
-  video              Video                    @relation(fields: [videoId], references: [aweme_id], onDelete: Cascade)
+  id                String                   @id @default(cuid())
+  videoId           String                   @unique
+  transcript        String[]
+  createdAt         DateTime                 @default(now())
+  updatedAt         DateTime                 @updatedAt
+  transcript_zh_tsv Unsupported("tsvector")?
+  video             Video                    @relation(fields: [videoId], references: [aweme_id], onDelete: Cascade)

  @@index([videoId])
  @@index([transcript_zh_tsv], map: "video_transcript_zh_idx", type: Gin)
 }
+
+model TranscriptionTask {
+  videoId        String    @id
+  status         String    @default("QUEUED")
+  providerTaskId String?
+  submittingAt   DateTime?
+  submittedAt    DateTime?
+  nextPollAt     DateTime  @default(now())
+  leaseUntil     DateTime?
+  leaseToken     String?
+  attempts       Int       @default(0)
+  lastError      String?
+  createdAt      DateTime  @default(now())
+  updatedAt      DateTime  @updatedAt
+  video          Video     @relation(fields: [videoId], references: [aweme_id], onDelete: Cascade)
+
+  @@index([status, nextPollAt])
+}