From 49547edf4224e2d7b0bb7140931a82bcbc028722 Mon Sep 17 00:00:00 2001
From: Brandon Thomas <echelon@gmail.com>
Date: Sat, 15 Oct 2016 23:32:16 -0400
Subject: [PATCH] 5-phones

---
 BUGS.md                 | 48 +++++++++++++++++++++++++++++++++++++++++
 FIX_PHONEMES.md         |  5 +++++
 dictionary/NEEDED.md    |  2 +-
 dictionary/cmudict-0.7b | 16 +++++++-------
 src/audiobank.rs        |  9 +++++++-
 src/synthesizer.rs      | 39 +++++++++++++++++++++++++++++++++
 6 files changed, 109 insertions(+), 10 deletions(-)
 create mode 100644 BUGS.md

diff --git a/BUGS.md b/BUGS.md
new file mode 100644
index 0000000000..357832af06
--- /dev/null
+++ b/BUGS.md
@@ -0,0 +1,48 @@
+Things that break
+=================
+
+On A.I.,
+and @GovGaryJohnson.
+RT @Cole_No_Slaw:
+Dems' plans
+Dem
+only $29.99!
+Thank you to our U.S. Navy
+ #TheGreaterGood.
+ at 2:00pm!
+ 60-66%
+140K
+ w/ corporate
+
+
+Times
+-----
+2:00pm        two p m
+7:30pm        seven thirty p m
+5:30ET        five thirty eastern time
+9pm ET        nine p m eastern time
+8pm ET        eight p m eastern time
+8:00 P.M.     eight p m
+9 p.m. ET     nine p m eastern time
+
+units:
+- AM, a.m.
+- et, pt, ET, PT
+
+2:00pm      -> 2:00 pm
+7:30p.m.    -> 7:30 p.m.
+9pm         -> 9 pm
+8:00pm      -> 8:00 pm
+5:30ET
+9pm ET
+8pm ET
+8:00 P.M.
+9 p.m. ET
+
+TimeWord
+  - am, a.m., AM, A.M.
+  - pm, p.m., PM, P.M.
+  - ET, PT
+
+\d:\d+\s?[(am)|(pm)]
+
diff --git a/FIX_PHONEMES.md b/FIX_PHONEMES.md
index 4fd77737bb..3307d50d72 100644
--- a/FIX_PHONEMES.md
+++ b/FIX_PHONEMES.md
@@ -2,3 +2,8 @@
 - K as in killing (K IH1 L IH0 NG)
 - IH1 as in killing (K IH1 L IH0 NG)
 - IH0 as in convict (K AA1 N V IH0 K T)
+
+Fix problems:
+-------------
+1M lives
+November 8, 
diff --git a/dictionary/NEEDED.md b/dictionary/NEEDED.md
index 17e758e9da..731a364818 100644
--- a/dictionary/NEEDED.md
+++ b/dictionary/NEEDED.md
@@ -22,11 +22,11 @@ patreon
 paypal
 pringles
 ramen
+sanic
 strudel
 sucky
 there's
 triforce
-troglodyte
 tweens
 typescript
 undressing
diff --git a/dictionary/cmudict-0.7b b/dictionary/cmudict-0.7b
index 65485a10e0..d722e5df73 100644
--- a/dictionary/cmudict-0.7b
+++ b/dictionary/cmudict-0.7b
@@ -9007,8 +9007,8 @@ BASING  B EY1 S IH0 NG
 BASINGER  B EY1 S IH0 NG ER0
 BASINS  B EY1 S AH0 N Z
 BASINSKI  B AH0 S IH1 N S K IY0
-BASIS  B EY1 S AH0 S
-BASIS(1)  B EY1 S IH0 S
+BASIS(1)  B EY1 S AH0 S
+BASIS  B EY1 S IH0 S
 BASISTA  B AH0 S IH1 S T AH0
 BASIX  B EY1 S IH0 K S
 BASK  B AE1 S K
@@ -92193,8 +92193,8 @@ PLACENTA  P L AH0 S EH1 N T AH0
 PLACENTIA  P L AH0 S EH1 N SH AH0
 PLACER  P L AE1 S ER0
 PLACER(1)  P L EY1 S ER0
-PLACES  P L EY1 S AH0 Z
-PLACES(1)  P L EY1 S IH0 Z
+PLACES(1)  P L EY1 S AH0 Z
+PLACES  P L EY1 S IH0 Z
 PLACETA  P L AH0 S EY1 T AH0
 PLACETAS  P L AH0 S EY1 T AH0 Z
 PLACEWAY  P L EY1 S W EY2
@@ -100571,8 +100571,8 @@ RESOURCEFULNESS  R IY0 S AO1 R S F AH0 L N AH0 S
 RESOURCES  R IY1 S AO0 R S IH0 Z
 RESOURCES'  R IY1 S AO0 R S IH0 Z
 RESPEAK  R IY2 S P IY1 K
-RESPECT  R IH0 S P EH1 K T
-RESPECT(1)  R IY0 S P EH1 K T
+RESPECT(1)  R IH0 S P EH1 K T
+RESPECT  R IY0 S P EH1 K T
 RESPECTABILITY  R IY0 S P EH2 K T AH0 B IH1 L IH0 T IY0
 RESPECTABLE  R IH0 S P EH1 K T AH0 B AH0 L
 RESPECTABLE(1)  R IY0 S P EH1 K T AH0 B AH0 L
@@ -107929,8 +107929,8 @@ SEMROW  S EH1 M R OW0
 SEMTEX  S EH1 M T EH0 K S
 SEN  S EH1 N
 SENA  S EH1 N AH0
-SENATE  S EH1 N AH0 T
-SENATE(1)  S EH1 N IH0 T
+SENATE(1)  S EH1 N AH0 T
+SENATE  S EH1 N IH0 T
 SENATE'S  S EH1 N IH0 T S
 SENATOR  S EH1 N AH0 T ER0
 SENATOR'S  S EH1 N AH0 T ER0 Z
diff --git a/src/audiobank.rs b/src/audiobank.rs
index c1daa71435..a6647bda23 100644
--- a/src/audiobank.rs
+++ b/src/audiobank.rs
@@ -134,8 +134,15 @@ impl Audiobank {
                   n_phone[0], n_phone[1], n_phone[2], n_phone[3]),
         )
       },
+      5 => {
+        (
+          "5-phones/",
+          format!("{}_{}_{}_{}_{}.wav",
+                  n_phone[0], n_phone[1], n_phone[2], n_phone[3], n_phone[4]),
+        )
+      },
       _ => {
-        return None; // Don't support more than 4-phones.
+        return None; // Don't support more than 5-phones.
       }
     };
 
diff --git a/src/synthesizer.rs b/src/synthesizer.rs
index 3c7742a708..65ee503bb8 100644
--- a/src/synthesizer.rs
+++ b/src/synthesizer.rs
@@ -352,6 +352,45 @@ impl Synthesizer {
       debug.push(None);
     }
 
+    // 5-phone
+    if polyphone.len() >= 5 {
+      let range = polyphone.len() - 4;
+      for i in 0..range {
+        if fulfilled[i]
+            || fulfilled[i+1]
+            || fulfilled[i+2]
+            || fulfilled[i+3] 
+            || fulfilled[i+4] {
+          continue;
+        }
+
+        let candidate_n_phone = &polyphone[i..i+5];
+
+        let sample_preference = match i {
+          0 => { SamplePreference::Begin },
+          _ if i == range - 1 => { SamplePreference::End },
+          _ => { SamplePreference::Middle },
+        };
+
+        //println!("n-phone: {:?}", candidate_n_phone);
+        let phone = self.audiobank.get_n_phone(speaker.as_str(),
+                                               candidate_n_phone,
+                                               sample_preference,
+                                               use_ends);
+
+        if phone.is_some() {
+          chunks[i] = phone;
+          debug[i] = Some(polyphone[i..i+5].join("_"));
+
+          for j in i..i+5 {
+            fulfilled[j] = true;
+          }
+        }
+      }
+    }
+
+    info!(target: "synthesis", "5-fulfilled: {:?}", fulfilled);
+
     // 4-phone
     if polyphone.len() >= 4 {
       let range = polyphone.len() - 3;