Update t2a.py

t2a.py

@@ -1,5 +1,7 @@
+import pydub
 from pydub import AudioSegment
 from tqdm import tqdm
+from pydub.playback import play
 import os
 from mappings import mappings, replacements
 
@@ -10,26 +12,46 @@ def text_to_audio(text, bpm):
     buffer_length = bpm_to_ms(bpm)
     audio = AudioSegment.silent(duration=0)
 
-    for key, value in replacements.items():
-        text = text.replace(key, value)
+    notes_list = text.split(" ")
+    for i in range(len(notes_list)):
+        note = notes_list[i]
+        # deduplicate characters in note
+        note = "".join(set(note))
+        notes_list[i] = note
 
-    for note in text.split(" "):
-        if note in mappings:
-            to_add = AudioSegment.from_wav(mappings[note])
-            # slice to be of consistent length or add more silence
-            if len(to_add) < buffer_length:
-                to_add = to_add + AudioSegment.silent(duration=buffer_length - len(to_add))
-            elif len(to_add) > buffer_length:
-                to_add = to_add[:buffer_length]
-            audio = audio + to_add
-        elif note == "n":
-            audio = audio + AudioSegment.silent(duration=buffer_length)
-        else: # everything else is a clap
-            to_add = AudioSegment.from_wav(mappings["l"])
-            # slice to be of consistent length or add more silence
-            if len(to_add) < buffer_length:
-                to_add = to_add + AudioSegment.silent(duration=buffer_length - len(to_add))
-            elif len(to_add) > buffer_length:
-                to_add = to_add[:buffer_length]
-            audio = audio + to_add
+    for note in notes_list:
+        to_add = None
+        for char in note:
+            if char in mappings:
+                if not to_add:
+                    to_add = AudioSegment.from_wav(mappings[char])
+                else:
+                    if char in mappings:
+                        new_audio = AudioSegment.from_wav(mappings[char])
+                        # get length of new_audio
+                        new_length = len(new_audio)
+                        # get length of to_add
+                        to_add_length = len(to_add)
+                        # if new_length is longer than to_add_length
+                        if new_length > to_add_length:
+                            # add silence to to_add
+                            to_add += AudioSegment.silent(duration=new_length - to_add_length)
+                        # if to_add_length is longer than new_length
+                        elif to_add_length > new_length:
+                            # add silence to new_audio
+                            new_audio += AudioSegment.silent(duration=to_add_length - new_length)
+                        
+                        to_add = to_add.overlay(new_audio)
+            elif char == "n":
+                to_add = AudioSegment.silent(duration=buffer_length)
+            else: # everything else is a clap
+                print('could not find mapping for ' + char)
+                to_add = AudioSegment.from_wav(mappings["l"])
+
+        if len(to_add) < buffer_length:
+            to_add = to_add + AudioSegment.silent(duration=buffer_length - len(to_add))
+        elif len(to_add) > buffer_length:
+            to_add = to_add[:buffer_length]
+        
+        audio = audio + to_add
     return audio