Add speaker-aware transcript schema

app/models/transcript.py

@@ -1,6 +1,20 @@
 from __future__ import annotations
 
-from pydantic import BaseModel, ConfigDict, Field
+from pydantic import BaseModel, ConfigDict, Field, model_validator
+
+
+class DiarizationSpan(BaseModel):
+    model_config = ConfigDict(extra="forbid")
+
+    speaker: str = Field(min_length=1, max_length=128)
+    start: float = Field(ge=0)
+    end: float = Field(ge=0)
+
+    @model_validator(mode="after")
+    def validate_range(self) -> "DiarizationSpan":
+        if self.end <= self.start:
+            raise ValueError("diarization span end must be greater than start")
+        return self
 
 
 class TranscriptWord(BaseModel):
@@ -9,6 +23,13 @@ class TranscriptWord(BaseModel):
     start: float = Field(ge=0)
     end: float = Field(ge=0)
     probability: float | None = Field(default=None, ge=0, le=1)
+    speaker: str | None = Field(default=None, min_length=1, max_length=128)
+
+    @model_validator(mode="after")
+    def validate_range(self) -> "TranscriptWord":
+        if self.end < self.start:
+            raise ValueError("word end must be greater than or equal to start")
+        return self
 
 
 class TranscriptSegment(BaseModel):
@@ -17,17 +38,25 @@ class TranscriptSegment(BaseModel):
     start: float = Field(ge=0)
     end: float = Field(ge=0)
     text: str
+    speaker: str | None = Field(default=None, min_length=1, max_length=128)
     words: list[TranscriptWord] = Field(default_factory=list)
 
+    @model_validator(mode="after")
+    def validate_range(self) -> "TranscriptSegment":
+        if self.end < self.start:
+            raise ValueError("segment end must be greater than or equal to start")
+        return self
+
 
 class Transcript(BaseModel):
     model_config = ConfigDict(extra="forbid")
-    schema_version: str = "1.0"
+    schema_version: str = "1.1"
     language: str
     language_probability: float = Field(ge=0, le=1)
     duration: float = Field(ge=0)
     task: str
     model: str
     text: str
+    speakers: list[str] = Field(default_factory=list)
     segments: list[TranscriptSegment] = Field(default_factory=list)
     word_timestamps: bool = True