diff --git a/tts_service/src/text.rs b/tts_service/src/text.rs
index d5aa98192e..b089a3e581 100644
--- a/tts_service/src/text.rs
+++ b/tts_service/src/text.rs
@@ -1,9 +1,9 @@
 use arpabet::Arpabet;
 use arpabet::phoneme::{Phoneme,VowelStress,Vowel,Consonant};
 use arpabet::Polyphone;
-use arpabet::extensions::Punctuation;
+use arpabet::extensions::Punctuation as ArpabetPunctuation;
 use arpabet::extensions::SentenceToken;
-use sentence::{SentenceTokenizer, Token};
+use sentence::{SentenceTokenizer, Token, Punctuation};
 use numerics::{NumericsBuilder, Numerics};
 use std::num::ParseIntError;
 
@@ -33,38 +33,85 @@ pub fn text_to_arpabet_encoding(arpabet: &Arpabet, text: &str) -> Vec<i64> {
   let text = text.to_ascii_lowercase();
 
   let tokens = sentence_tokenizer.tokenize(&text);
+  println!("Sentence Tokens: {:?}", tokens);
 
-  let space = u8::from(Punctuation::Space) as i64;
-  let start= u8::from(Punctuation::StartToken) as i64;
-  let end = u8::from(Punctuation::EndToken) as i64;
+  // Token encodings
+  let space = u8::from(ArpabetPunctuation::Space) as i64;
+  let start= u8::from(ArpabetPunctuation::StartToken) as i64;
+  let end = u8::from(ArpabetPunctuation::EndToken) as i64;
+
+  // TODO: These don't sound great.
+  let question = u8::from(ArpabetPunctuation::Question) as i64;
+  let period = u8::from(ArpabetPunctuation::Period) as i64;
+  let exclamation = u8::from(ArpabetPunctuation::Exclamation) as i64;
+  let comma = u8::from(ArpabetPunctuation::Comma) as i64;
+  let interjection = u8::from(ArpabetPunctuation::Interjection) as i64;
 
   let mut encoded_buffer : Vec<i64> = Vec::new();
   encoded_buffer.push(start); // start token
 
+  let mut needs_space = false;
+
   for token in tokens {
     match token {
       Token::Word(word) => {
-        if arpabet_lookup_and_encode_word(&word, &arpabet, &mut encoded_buffer) {
+        if needs_space {
           encoded_buffer.push(space);
+          needs_space = false;
+        }
+        if arpabet_lookup_and_encode_word(&word, &arpabet, &mut encoded_buffer) {
+          needs_space = true;
         }
       },
       Token::Integer(integer_string) => {
+        if needs_space {
+          encoded_buffer.push(space);
+          needs_space = false;
+        }
         let number_words = integer_string.parse::<i64>()
             .ok()
             .and_then(|number| numerics.convert_number(number).ok())
             .unwrap_or(Vec::new());
 
         for word in number_words {
-          if arpabet_lookup_and_encode_word(&word, &arpabet, &mut encoded_buffer) {
+          if needs_space {
             encoded_buffer.push(space);
+            needs_space = false;
+          }
+          if arpabet_lookup_and_encode_word(&word, &arpabet, &mut encoded_buffer) {
+            needs_space = true;
           }
         }
       },
-      Token::Punctuation(_) => continue,
+      Token::Punctuation(punctuation) => {
+        match punctuation {
+          Punctuation::Period => {
+            encoded_buffer.push(period);
+            needs_space = true;
+          },
+          Punctuation::Comma | Punctuation::Semicolon => {
+            encoded_buffer.push(comma);
+            needs_space = true;
+          },
+          Punctuation::Question => {
+            encoded_buffer.push(question);
+            needs_space = true;
+          },
+          Punctuation::Exclamation => {
+            encoded_buffer.push(exclamation);
+            needs_space = true;
+          },
+          Punctuation::Colon | Punctuation::Dash => {
+            encoded_buffer.push(interjection);
+            needs_space = true;
+          },
+        }
+      },
+      // TODO:
+      Token::RealNumber(_) => continue,
       Token::CommaFormattedInteger(_) => continue,
       Token::CommaFormattedRealNumber(_) => continue,
       Token::Hashtag(_) => continue,
-      Token::RealNumber(_) => continue,
       Token::UsernameMention(_) => continue,
       Token::Url(_) => continue,
       Token::Unknown(_) => continue, // skip, we don't know what this is yet.