legofan94 commited on
Commit
abd10df
1 Parent(s): 3d2d856

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +14 -5
app.py CHANGED
@@ -41,7 +41,11 @@ def generate_roast(image_path):
41
  model = genai.GenerativeModel(
42
  model_name="gemini-1.5-flash-002",
43
  generation_config=generation_config,
44
- system_instruction="You are a professional satirist and fashion expert. Roast the provided profile picture in less than 50 words.",
 
 
 
 
45
  )
46
 
47
  chat_session = model.start_chat(
@@ -55,14 +59,19 @@ def generate_roast(image_path):
55
  # Function to convert text to speech with Play.ht
56
  def text_to_speech(text):
57
  try:
58
- url = "https://api.play.ht/api/v2/tts/stream"
59
  payload = {
60
- "voice": "s3://voice-cloning-zero-shot/d9ff78ba-d016-47f6-b0ef-dd630f59414e/female-cs/manifest.json",
 
 
 
 
 
 
61
  "output_format": "mp3",
62
- "text": text,
63
  }
64
  headers = {
65
- "accept": "audio/mpeg",
66
  "content-type": "application/json",
67
  "Authorization": API_KEY,
68
  "X-User-ID": USER_ID
 
41
  model = genai.GenerativeModel(
42
  model_name="gemini-1.5-flash-002",
43
  generation_config=generation_config,
44
+ system_instruction="Generate a conversation between two women gently roasting the uploaded image in less than 100 words. Please abide by these guidelines. \
45
+ 1. Begin conversation turns with the prefix 'Host: 1' and 'Host: 2'. For example, Host 1: Hello how are you? Host 2: I'm good and yourself? Host 3: Thanks for asking! \
46
+ 2. Use humor, irony, and sarcasm to gently roast the picture \
47
+ 3. Your output should be a well-written text suitable for reading aloud. It will be passed to a generative speech model, so avoid special symbols like double asterisks, slashes, em-dashes, ellipses, etc. Also avoid output that isn't dialogue. \
48
+ 4. Conversation turns should be short and snappy",
49
  )
50
 
51
  chat_session = model.start_chat(
 
59
  # Function to convert text to speech with Play.ht
60
  def text_to_speech(text):
61
  try:
62
+ url = "https://api.play.ai/api/v1/tts/stream"
63
  payload = {
64
+ "model": "PlayDialog",
65
+ "voice": "s3://voice-cloning-zero-shot/adb83b67-8d75-48ff-ad4d-a0840d231ef1/original/manifest.json",
66
+ "voice2": "s3://voice-cloning-zero-shot/50381567-ff7b-46d2-bfdc-a9584a85e08d/original/manifest.json",
67
+ "turnPrefix": "Host 1:",
68
+ "turnPrefix2": "Host 2:",
69
+ 'prompt': None,
70
+ 'prompt2': None,
71
  "output_format": "mp3",
72
+ "text": text,
73
  }
74
  headers = {
 
75
  "content-type": "application/json",
76
  "Authorization": API_KEY,
77
  "X-User-ID": USER_ID