diff --git a/src/assistant/lib/tts/__init__.py b/src/assistant/lib/tts/__init__.py
new file mode 100644
index 0000000..e69de29
diff --git a/src/assistant/lib/tts/models/__init__.py b/src/assistant/lib/tts/models/__init__.py
new file mode 100644
index 0000000..f3a089c
--- /dev/null
+++ b/src/assistant/lib/tts/models/__init__.py
@@ -0,0 +1,11 @@
+from .models import *
+from .voice import *
+
+__all__ = [
+    "LanguageCodes",
+    "TTSRequestModel",
+    "TTSResponseModel",
+    "VoiceModel",
+    "YandexVoiceModelNamesString",
+    "YandexVoiceModels",
+]
diff --git a/src/assistant/lib/tts/models/models.py b/src/assistant/lib/tts/models/models.py
new file mode 100644
index 0000000..14ee677
--- /dev/null
+++ b/src/assistant/lib/tts/models/models.py
@@ -0,0 +1,13 @@
+import pydantic
+
+import lib.tts.models.voice as tts_models_voice
+
+
+class TTSRequestModel(pydantic.BaseModel):
+    model_config = pydantic.ConfigDict(use_enum_values=True)
+
+    voice_model_name: tts_models_voice.YandexVoiceModelNamesString
+
+
+class TTSResponseModel(pydantic.BaseModel):
+    audio_content: bytes
diff --git a/src/assistant/lib/tts/models/voice/__init__.py b/src/assistant/lib/tts/models/voice/__init__.py
new file mode 100644
index 0000000..ce1a9cf
--- /dev/null
+++ b/src/assistant/lib/tts/models/voice/__init__.py
@@ -0,0 +1,9 @@
+from .base import *
+from .yandex import *
+
+__all__ = [
+    "LanguageCodes",
+    "VoiceModel",
+    "YandexVoiceModelNamesString",
+    "YandexVoiceModels",
+]
diff --git a/src/assistant/lib/tts/models/voice/base.py b/src/assistant/lib/tts/models/voice/base.py
new file mode 100644
index 0000000..65082f1
--- /dev/null
+++ b/src/assistant/lib/tts/models/voice/base.py
@@ -0,0 +1,20 @@
+import enum
+
+import pydantic
+
+
+class LanguageCodes(enum.Enum):
+    RUSSIAN = "ru-RU"
+    ENGLISH = "en-US"
+    KAZAKH = "kk-KK"
+    GERMAN = "de-DE"
+    HEBREW = "he-IL"
+    UZBEK = "uz-UZ"
+
+
+class VoiceModel(pydantic.BaseModel):
+    model_config = pydantic.ConfigDict(use_enum_values=True)
+
+    voice_name: str
+    role: str | None = None
+    lang: LanguageCodes
diff --git a/src/assistant/lib/tts/models/voice/yandex.py b/src/assistant/lib/tts/models/voice/yandex.py
new file mode 100644
index 0000000..3a47c6b
--- /dev/null
+++ b/src/assistant/lib/tts/models/voice/yandex.py
@@ -0,0 +1,126 @@
+import enum
+
+import lib.tts.models.voice.base as tts_models_voice_base
+
+
+class YandexVoiceModelNamesString(enum.Enum):
+    ERMIL_NEUTRAL = "ermil neutral"
+    ERMIL_GOOD = "ermil good"
+    ALENA_NEUTRAL = "alena neutral"
+    ALENA_GOOD = "alena good"
+    JANE_NEUTRAL = "jane neutral"
+    JANE_GOOD = "jane good"
+    JANE_EVIL = "jane evil"
+    OMAZH_NEUTRAL = "omazh neutral"
+    OMAZH_EVIL = "omazh evil"
+    ZAHAR_NEUTRAL = "zahar neutral"
+    ZAHAR_GOOD = "zahar good"
+    FILIPP = "filipp"
+    MADIRUS = "madirus"
+    DASHA = "dasha"
+    JULIA = "julia"
+    LERA = "lera"
+    MARINA = "marina"
+    ALEXANDER = "alexander"
+    KIRILL = "kirill"
+    ANTON = "anton"
+    # English
+    JOHN = "john"
+    # Kazakh
+    AMIRA = "amira"
+    MADI = "madi"
+    # German
+    LEA = "lea"
+    # HEBREW
+    NAOMI_MODERN = "naomi modern"
+    NAOMI_CLASSIC = "naomi classic"
+    # Uzbek
+    NIGORA = "nigora"
+
+
+class YandexVoiceModels(enum.Enum):
+    # Russian
+    ERMIL_NEUTRAL = tts_models_voice_base.VoiceModel(
+        voice_name="ermil", role="neutral", lang=tts_models_voice_base.LanguageCodes.RUSSIAN
+    )
+    ERMIL_GOOD = tts_models_voice_base.VoiceModel(
+        voice_name="ermil", role="good", lang=tts_models_voice_base.LanguageCodes.RUSSIAN
+    )
+    ALENA_NEUTRAL = tts_models_voice_base.VoiceModel(
+        voice_name="alena", role="neutral", lang=tts_models_voice_base.LanguageCodes.RUSSIAN
+    )
+    ALENA_GOOD = tts_models_voice_base.VoiceModel(
+        voice_name="alena", role="good", lang=tts_models_voice_base.LanguageCodes.RUSSIAN
+    )
+    JANE_NEUTRAL = tts_models_voice_base.VoiceModel(
+        voice_name="jane", role="neutral", lang=tts_models_voice_base.LanguageCodes.RUSSIAN
+    )
+    JANE_GOOD = tts_models_voice_base.VoiceModel(
+        voice_name="jane", role="good", lang=tts_models_voice_base.LanguageCodes.RUSSIAN
+    )
+    JANE_EVIL = tts_models_voice_base.VoiceModel(
+        voice_name="jane", role="evil", lang=tts_models_voice_base.LanguageCodes.RUSSIAN
+    )
+    OMAZH_NEUTRAL = tts_models_voice_base.VoiceModel(
+        voice_name="omazh", role="neutral", lang=tts_models_voice_base.LanguageCodes.RUSSIAN
+    )
+    OMAZH_EVIL = tts_models_voice_base.VoiceModel(
+        voice_name="omazh", role="evil", lang=tts_models_voice_base.LanguageCodes.RUSSIAN
+    )
+    ZAHAR_NEUTRAL = tts_models_voice_base.VoiceModel(
+        voice_name="zahar", role="neutral", lang=tts_models_voice_base.LanguageCodes.RUSSIAN
+    )
+    ZAHAR_GOOD = tts_models_voice_base.VoiceModel(
+        voice_name="zahar", role="good", lang=tts_models_voice_base.LanguageCodes.RUSSIAN
+    )
+    FILIPP = tts_models_voice_base.VoiceModel(
+        voice_name="filipp", role=None, lang=tts_models_voice_base.LanguageCodes.RUSSIAN
+    )
+    MADIRUS = tts_models_voice_base.VoiceModel(
+        voice_name="madirus", role=None, lang=tts_models_voice_base.LanguageCodes.RUSSIAN
+    )
+    DASHA = tts_models_voice_base.VoiceModel(
+        voice_name="dasha", role=None, lang=tts_models_voice_base.LanguageCodes.RUSSIAN
+    )
+    JULIA = tts_models_voice_base.VoiceModel(
+        voice_name="julia", role=None, lang=tts_models_voice_base.LanguageCodes.RUSSIAN
+    )
+    LERA = tts_models_voice_base.VoiceModel(
+        voice_name="lera", role=None, lang=tts_models_voice_base.LanguageCodes.RUSSIAN
+    )
+    MARINA = tts_models_voice_base.VoiceModel(
+        voice_name="marina", role=None, lang=tts_models_voice_base.LanguageCodes.RUSSIAN
+    )
+    ALEXANDER = tts_models_voice_base.VoiceModel(
+        voice_name="alexander", role=None, lang=tts_models_voice_base.LanguageCodes.RUSSIAN
+    )
+    KIRILL = tts_models_voice_base.VoiceModel(
+        voice_name="kirill", role=None, lang=tts_models_voice_base.LanguageCodes.RUSSIAN
+    )
+    ANTON = tts_models_voice_base.VoiceModel(
+        voice_name="anton", role=None, lang=tts_models_voice_base.LanguageCodes.RUSSIAN
+    )
+    # English
+    JOHN = tts_models_voice_base.VoiceModel(
+        voice_name="john", role=None, lang=tts_models_voice_base.LanguageCodes.ENGLISH
+    )
+    # Kazakh
+    AMIRA = tts_models_voice_base.VoiceModel(
+        voice_name="amira", role=None, lang=tts_models_voice_base.LanguageCodes.KAZAKH
+    )
+    MADI = tts_models_voice_base.VoiceModel(
+        voice_name="madi", role=None, lang=tts_models_voice_base.LanguageCodes.KAZAKH
+    )
+    # German
+    LEA = tts_models_voice_base.VoiceModel(voice_name="lea", role=None, lang=tts_models_voice_base.LanguageCodes.GERMAN)
+    # HEBREW
+    NAOMI_MODERN = tts_models_voice_base.VoiceModel(
+        voice_name="naomi", role="modern", lang=tts_models_voice_base.LanguageCodes.HEBREW
+    )
+    NAOMI_CLASSIC = tts_models_voice_base.VoiceModel(
+        voice_name="naomi", role="classic", lang=tts_models_voice_base.LanguageCodes.HEBREW
+    )
+    # Uzbek
+    NIGORA = tts_models_voice_base.VoiceModel(
+        voice_name="nigora", role=None, lang=tts_models_voice_base.LanguageCodes.UZBEK
+    )