Add Contextual Information to Responses (#11)

* Save conversation to text.

* Allow users to customize the prompts and get contextual responses.

* Update readme with feature request for Contextual information.

* Update button text.

AudioRecorder.py

@@ -6,6 +6,7 @@
 ENERGY_THRESHOLD = 1000
 DYNAMIC_ENERGY_THRESHOLD = False
 
+
 class BaseRecorder:
  def __init__(self, source, source_name):
  self.recorder = sr.Recognizer()
@@ -31,11 +32,13 @@ def record_callback(_, audio:sr.AudioData) -> None:
 
  self.recorder.listen_in_background(self.source, record_callback, phrase_time_limit=RECORD_TIMEOUT)
 
+
 class DefaultMicRecorder(BaseRecorder):
  def __init__(self):
  super().__init__(source=sr.Microphone(sample_rate=16000), source_name="You")
  self.adjust_for_noise("Default Mic", "Please make some noise from the Default Mic...")
 
+
 class DefaultSpeakerRecorder(BaseRecorder):
  def __init__(self):
  with pyaudio.PyAudio() as p:

CustomPrompts.py

@@ -0,0 +1,19 @@
+# A user can customize responses based on contextual information.
+# E.g. A biologist might want to provide answers specific to certain study papers.
+# An advanced programmer might want to provide answers specific to programming in Rust
+#
+# It requires some knowledge of Prompt Engineering to craft good preamble, epilogues
+
+DEFAULT_PREAMBLE = """You are a casual pal, genuinely interested in the conversation at hand.""" \
+ """A poor transcription of conversation is given below."""
+
+DEFAULT_EPILOGUE = """Please respond, in detail, to the conversation. Confidently give a """\
+ """straightforward response to the speaker, even if you don't understand """\
+ """them. Give your response in square brackets. DO NOT ask to repeat, """\
+ """and DO NOT ask for clarification. Just answer the speaker directly."""
+
+# To provide custom preamble, epilogue, define new constants and assign them to PREAMBLE, EPILOGUE
+# Both preamble and epilogue are required
+
+PREAMBLE = DEFAULT_PREAMBLE
+EPILOGUE = DEFAULT_EPILOGUE

GPTResponder.py

@@ -7,11 +7,13 @@
 # Number of phrases to use for generating a response
 MAX_PHRASES = 10
 
+
 def generate_response_from_transcript(transcript):
  try:
+ prompt_content = create_prompt(transcript)
  response = openai.ChatCompletion.create(
  model="gpt-3.5-turbo-0301",
- messages=[{"role": "system", "content": create_prompt(transcript)}],
+ messages=[{"role": "system", "content": prompt_content}],
  temperature=0.0
  )
  except Exception as exception:

README.md

@@ -102,6 +102,7 @@ This project is licensed under the MIT License - see the [LICENSE](LICENSE) file
 
 ## ➕ Enhancements from base repository ➕
 - Do not need Open AI key, paid Open AI account to use the complete functionality
+- Allow contexttual information to provide customized responses to users
 - Transcribe any video
 - Preserve all conversation text in UI
 - Allow saving conversation to file

custom_speech_recognition/__init__.py

@@ -36,7 +36,7 @@
 from .audio import AudioData, get_flac_converter
 from .exceptions import (
  RequestError,
- TranscriptionFailed, 
+ TranscriptionFailed,
  TranscriptionNotReady,
  UnknownValueError,
  WaitTimeoutError,

main.py

@@ -79,7 +79,7 @@ def create_ui_components(root):
  update_interval_slider.set(2)
  update_interval_slider.grid(row=3, column=1, padx=10, pady=10, sticky="nsew")
 
- copy_button = ctk.CTkButton(root, text="Copy", command=None)
+ copy_button = ctk.CTkButton(root, text="Copy Audio Transcript", command=None)
  copy_button.grid(row=2, column=0, padx=10, pady=3, sticky="nsew")
 
  save_file_button = ctk.CTkButton(root, text="Save to File", command=None)
@@ -106,8 +106,7 @@ def main():
  '\nbase model has to be downloaded from the link https://drive.google.com/file/d/1E44DVjpfZX8tSrSagaDJXU91caZOkwa6/view?usp=drive_link'
  '\nsmall model has to be downloaded from the link https://drive.google.com/file/d/1E44DVjpfZX8tSrSagaDJXU91caZOkwa6/view?usp=drive_link'
  '\nOpenAI has more models besides the ones specified above.'
- '\nThose models are prohibitive to use on local machines because of memory requirements.'
- '\nThis option is only applicable when not using the --api option.')
+ '\nThose models are prohibitive to use on local machines because of memory requirements.')
  args = cmd_args.parse_args()
 
  try:

prompts.py

@@ -1,12 +1,10 @@
-INITIAL_RESPONSE = "Welcome to Transcribe 🤝 👋"
+from CustomPrompts import PREAMBLE, EPILOGUE
 
+INITIAL_RESPONSE = '👋 Welcome to Transcribe 🤝'
 
-def create_prompt(transcript):
- return f"""You are a casual pal, genuinely interested in the conversation at hand. \
- A poor transcription of conversation is given below. 
-
-{transcript}.
 
-Please respond, in detail, to the conversation. Confidently give a straightforward response to the speaker, 
-even if you don't understand them. Give your response in square brackets. DO NOT ask to repeat, and DO NOT ask for 
-clarification. Just answer the speaker directly."""
+def create_prompt(transcript):
+ return f'{PREAMBLE} \
+ \
+{transcript}.\
+{EPILOGUE}'