diff --git a/homeassistant/components/stt/__init__.py b/homeassistant/components/stt/__init__.py
index 679f9b29e41..b1730a09357 100644
--- a/homeassistant/components/stt/__init__.py
+++ b/homeassistant/components/stt/__init__.py
@@ -40,12 +40,11 @@ from .const import (
 )
 from .legacy import (
     Provider,
-    SpeechMetadata,
-    SpeechResult,
     async_default_provider,
     async_get_provider,
     async_setup_legacy,
 )
+from .models import SpeechMetadata, SpeechResult
 
 __all__ = [
     "async_get_provider",
diff --git a/homeassistant/components/stt/legacy.py b/homeassistant/components/stt/legacy.py
index f14eed467db..862f59d5f6d 100644
--- a/homeassistant/components/stt/legacy.py
+++ b/homeassistant/components/stt/legacy.py
@@ -3,7 +3,6 @@ from __future__ import annotations
 
 from abc import ABC, abstractmethod
 from collections.abc import AsyncIterable, Coroutine
-from dataclasses import dataclass
 import logging
 from typing import Any
 
@@ -20,8 +19,8 @@ from .const import (
     AudioCodecs,
     AudioFormats,
     AudioSampleRates,
-    SpeechResultState,
 )
+from .models import SpeechMetadata, SpeechResult
 
 _LOGGER = logging.getLogger(__name__)
 
@@ -88,32 +87,6 @@ def async_setup_legacy(
     ]
 
 
-@dataclass
-class SpeechMetadata:
-    """Metadata of audio stream."""
-
-    language: str
-    format: AudioFormats
-    codec: AudioCodecs
-    bit_rate: AudioBitRates
-    sample_rate: AudioSampleRates
-    channel: AudioChannels
-
-    def __post_init__(self) -> None:
-        """Finish initializing the metadata."""
-        self.bit_rate = AudioBitRates(int(self.bit_rate))
-        self.sample_rate = AudioSampleRates(int(self.sample_rate))
-        self.channel = AudioChannels(int(self.channel))
-
-
-@dataclass
-class SpeechResult:
-    """Result of audio Speech."""
-
-    text: str | None
-    result: SpeechResultState
-
-
 class Provider(ABC):
     """Represent a single STT provider."""
 
diff --git a/homeassistant/components/stt/models.py b/homeassistant/components/stt/models.py
new file mode 100644
index 00000000000..45322e2da07
--- /dev/null
+++ b/homeassistant/components/stt/models.py
@@ -0,0 +1,37 @@
+"""Speech-to-text data models."""
+from dataclasses import dataclass
+
+from .const import (
+    AudioBitRates,
+    AudioChannels,
+    AudioCodecs,
+    AudioFormats,
+    AudioSampleRates,
+    SpeechResultState,
+)
+
+
+@dataclass
+class SpeechMetadata:
+    """Metadata of audio stream."""
+
+    language: str
+    format: AudioFormats
+    codec: AudioCodecs
+    bit_rate: AudioBitRates
+    sample_rate: AudioSampleRates
+    channel: AudioChannels
+
+    def __post_init__(self) -> None:
+        """Finish initializing the metadata."""
+        self.bit_rate = AudioBitRates(int(self.bit_rate))
+        self.sample_rate = AudioSampleRates(int(self.sample_rate))
+        self.channel = AudioChannels(int(self.channel))
+
+
+@dataclass
+class SpeechResult:
+    """Result of audio Speech."""
+
+    text: str | None
+    result: SpeechResultState