update eleven labs to have premade voices by default

earlier, generated voices from the voice labs were displayed. another auxiliary change is made in this commit: - the API for eleven labs now accepts elevenlabs_voice_name rather than elevenlabs_voice_id. this makes the API easier to use.
GooeyAI · Oct 6, 2023 · 370b2f5 · 370b2f5
1 parent ab6ad30
commit 370b2f5
Show file tree

Hide file tree

Showing 2 changed files with 50 additions and 15 deletions.
diff --git a/daras_ai_v2/text_to_speech_settings_widgets.py b/daras_ai_v2/text_to_speech_settings_widgets.py
@@ -27,13 +27,47 @@ class TextToSpeechProviders(Enum):
     ELEVEN_LABS = "ElevenLabs"
 
 
-# Mapping from Eleven Labs Voice ID -> Title in UI
+# Mapping from Eleven Labs Voice Name -> Voice ID
 ELEVEN_LABS_VOICES = {
-    "syN7Wt0nfLXAqq9LI9R6": "Indian Man With A Deep Voice",
-    "99XLBDkANYZ0Ww7MKN8d": "[aaa] Ramakrishnan - INDIAN TEACHER. CEREBRAL AND THOUGHTFUL PRONUNCIATION 🔥",
-    "TSh1KthMgVjY3BfYFkwS": "[ElevenVoices] Riya - Indian Female Young Adult",
-    "dvRCseVM0rT31i8clSVy": "Wise Grandma  slow, seductive, pleasant, mature",
-    "prqKmUi0Zo7WBmA81Vy4": "[ElevenVoices] Rahul - Indian Male Young Adult",
+  "Rachel": "21m00Tcm4TlvDq8ikWAM",
+  "Clyde": "2EiwWnXFnvU5JabPnv8n",
+  "Domi": "AZnzlk1XvdvUeBnXmlld",
+  "Dave": "CYw3kZ02Hs0563khs1Fj",
+  "Fin": "D38z5RcWu1voky8WS1ja",
+  "Bella": "EXAVITQu4vr4xnSDxMaL",
+  "Antoni": "ErXwobaYiN019PkySvjV",
+  "Thomas": "GBv7mTt0atIp3Br8iCZE",
+  "Charlie": "IKne3meq5aSn9XLyUdCD",
+  "Emily": "LcfcDJNUP1GQjkzn1xUU",
+  "Elli": "MF3mGyEYCl7XYWbV9V6O",
+  "Callum": "N2lVS1w4EtoT3dr4eOWO",
+  "Patrick": "ODq5zmih8GrVes37Dizd",
+  "Harry": "SOYHLrjzK2X1ezoPC6cr",
+  "Liam": "TX3LPaxmHKxFdv7VOQHJ",
+  "Dorothy": "ThT5KcBeYPX3keUQqHPh",
+  "Josh": "TxGEqnHWrfWFTfGW9XjX",
+  "Arnold": "VR6AewLTigWG4xSOukaG",
+  "Charlotte": "XB0fDUnXU5powFXDhCwa",
+  "Matilda": "XrExE9yKIg1WjnnlVkGX",
+  "Matthew": "Yko7PKHZNXotIFUBG7I9",
+  "James": "ZQe5CZNOzWyzPSCn5a3c",
+  "Joseph": "Zlb1dXrM653N07WRdFW3",
+  "Jeremy": "bVMeCyTHy58xNoL34h3p",
+  "Michael": "flq6f7yk4E4fJM5XTYuZ",
+  "Ethan": "g5CIjZEefAph4nQFvHAz",
+  "Gigi": "jBpfuIE2acCO8z3wKNLl",
+  "Freya": "jsCqWAovK2LkecY7zXl4",
+  "Grace": "oWAxZDx7w5VEj9dCyTzz",
+  "Daniel": "onwK4e9ZLuTAKqWW03F9",
+  "Serena": "pMsXgVXv3BLzUgSXRplE",
+  "Adam": "pNInz6obpgDQGcFmaJgB",
+  "Nicole": "piTKgcLEGmPE4e6mEKli",
+  "Jessie": "t0jbNlBVZ17f02VDIeMI",
+  "Ryan": "wViXBPUzp2ZZixB1xQuM",
+  "Sam": "yoZ06aMxZJJ28mfd3POQ",
+  "Glinda": "z9fAnlkpzviPz146aGWa",
+  "Giovanni": "zcAOhNBS3c14rBihAFp1",
+  "Mimi": "zrHiDhphv9ZnVXBqCLjz",
 }
 
 # Mapping from Model ID -> Title in UI
@@ -165,7 +199,7 @@ def text_to_speech_settings():
                     ###### Voice name (ElevenLabs)
                     """,
                     key="elevenlabs_voice_name",
-                    format_func=ELEVEN_LABS_VOICES.__getitem__,
+                    format_func=str,
                     options=ELEVEN_LABS_VOICES.keys(),
                 )
 

diff --git a/recipes/TextToSpeech.py b/recipes/TextToSpeech.py
@@ -42,7 +42,7 @@ class TextToSpeechPage(BasePage):
         "google_speaking_rate": 1.0,
         "uberduck_voice_name": "Aiden Botha",
         "uberduck_speaking_rate": 1.0,
-        "elevenlabs_voice_name": "syN7Wt0nfLXAqq9LI9R6",
+        "elevenlabs_voice_name": "Rachel",
         "elevenlabs_model": "eleven_multilingual_v2",
         "elevenlabs_stability": 0.5,
         "elevenlabs_similarity_boost": 0.75,
@@ -251,19 +251,20 @@ def run(self, state: dict):
                 )
 
             case TextToSpeechProviders.ELEVEN_LABS:
-                # default to first voice ID in the mapping
-                default_voice_id = next(iter(ELEVEN_LABS_VOICES))
-
-                # default to first model ID in the mapping
+                # default to first in the mapping
                 default_voice_model = next(iter(ELEVEN_LABS_MODELS))
+                default_voice_name = next(iter(ELEVEN_LABS_VOICES))
 
-                voice_id = state.get("elevenlabs_voice_name", default_voice_id)
                 voice_model = state.get("elevenlabs_model", default_voice_model)
+                voice_name = state.get("elevenlabs_voice_name", default_voice_name)
 
-                if voice_id not in ELEVEN_LABS_VOICES:
-                    raise ValueError(f"Invalid voice_name: {voice_id}")
+                # validate voice_model / voice_name
                 if voice_model not in ELEVEN_LABS_MODELS:
                     raise ValueError(f"Invalid model: {voice_model}")
+                if voice_name not in ELEVEN_LABS_VOICES:
+                    raise ValueError(f"Invalid voice_name: {voice_name}")
+                else:
+                    voice_id = ELEVEN_LABS_VOICES[voice_name]
 
                 stability = state.get("elevenlabs_stability", 0.5)
                 similarity_boost = state.get("elevenlabs_similarity_boost", 0.75)