diff --git a/homeassistant/components/amazon_polly/tts.py b/homeassistant/components/amazon_polly/tts.py
index 3492518421d..d1c12e657fd 100644
--- a/homeassistant/components/amazon_polly/tts.py
+++ b/homeassistant/components/amazon_polly/tts.py
@@ -2,6 +2,7 @@
 import logging
 
 import boto3
+import botocore
 import voluptuous as vol
 
 from homeassistant.components.tts import PLATFORM_SCHEMA, Provider
@@ -125,6 +126,10 @@ DEFAULT_TEXT_TYPE = "text"
 
 DEFAULT_SAMPLE_RATES = {"mp3": "22050", "ogg_vorbis": "22050", "pcm": "16000"}
 
+AWS_CONF_CONNECT_TIMEOUT = 10
+AWS_CONF_READ_TIMEOUT = 5
+AWS_CONF_MAX_POOL_CONNECTIONS = 1
+
 PLATFORM_SCHEMA = PLATFORM_SCHEMA.extend(
     {
         vol.Optional(CONF_REGION, default=DEFAULT_REGION): vol.In(SUPPORTED_REGIONS),
@@ -167,6 +172,11 @@ def get_engine(hass, config, discovery_info=None):
         CONF_REGION: config[CONF_REGION],
         CONF_ACCESS_KEY_ID: config.get(CONF_ACCESS_KEY_ID),
         CONF_SECRET_ACCESS_KEY: config.get(CONF_SECRET_ACCESS_KEY),
+        "config": botocore.config.Config(
+            connect_timeout=AWS_CONF_CONNECT_TIMEOUT,
+            read_timeout=AWS_CONF_READ_TIMEOUT,
+            max_pool_connections=AWS_CONF_MAX_POOL_CONNECTIONS,
+        ),
     }
 
     del config[CONF_REGION]
@@ -229,6 +239,7 @@ class AmazonPollyProvider(Provider):
             _LOGGER.error("%s does not support the %s language", voice_id, language)
             return None, None
 
+        _LOGGER.debug("Requesting TTS file for text: %s", message)
         resp = self.client.synthesize_speech(
             Engine=self.config[CONF_ENGINE],
             OutputFormat=self.config[CONF_OUTPUT_FORMAT],
@@ -238,6 +249,7 @@ class AmazonPollyProvider(Provider):
             VoiceId=voice_id,
         )
 
+        _LOGGER.debug("Reply received for TTS: %s", message)
         return (
             CONTENT_TYPE_EXTENSIONS[resp.get("ContentType")],
             resp.get("AudioStream").read(),