vikarti.anatra/AudioSumma
1
0
Fork 0
mirror of https://github.com/patw/AudioSumma.git synced 2025-09-04 11:40:34 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Switched bytes to tokens to be consistent with output. Allowed much larger outputs and inputs to support long context models.

Browse source
This commit is contained in:
Pat Wendorf 2025-07-02 12:50:24 -04:00
parent 6e2fa21b62
commit 9b8a6077bf
1 changed files with 7 additions and 6 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

13
meetings.py
View file

@ -35,7 +35,7 @@ class RecordingApp(QWidget):
"SUMMARY_PROMPT": "Call Transcript: {chunk}\n\nInstruction: Summarize the above call transcript but DO NOT MENTION THE TRANSCRIPT", "SUMMARY_PROMPT": "Call Transcript: {chunk}\n\nInstruction: Summarize the above call transcript but DO NOT MENTION THE TRANSCRIPT",
"FACT_PROMPT": "Call Transcript: {chunk}\n\nInstruction: Summarize all the facts in the transcript, one per line bullet point", "FACT_PROMPT": "Call Transcript: {chunk}\n\nInstruction: Summarize all the facts in the transcript, one per line bullet point",
"SENTIMENT_PROMPT": "Call Transcript: {chunk}\n\nInstruction: Summarize the sentiment for topics in the above call transcript but DO NOT MENTION THE TRANSCRIPT", "SENTIMENT_PROMPT": "Call Transcript: {chunk}\n\nInstruction: Summarize the sentiment for topics in the above call transcript but DO NOT MENTION THE TRANSCRIPT",
"CHUNK_SIZE": 12288, "CHUNK_SIZE": 3000,
"MAX_TOKENS": 1000 "MAX_TOKENS": 1000
} }
@ -228,9 +228,10 @@ class RecordingApp(QWidget):
chunks = [] chunks = []
lines = string.split("\n") lines = string.split("\n")
current_chunk = "" current_chunk = ""
chunk_bytes = chunk_size * 4 # Tokens average out to 4 bytes each
for line in lines: for line in lines:
current_chunk += line current_chunk += line
if len(current_chunk) >= chunk_size: if len(current_chunk) >= chunk_bytes:
chunks.append(current_chunk) chunks.append(current_chunk)
current_chunk = "" current_chunk = ""
if current_chunk: if current_chunk:
@ -330,10 +331,10 @@ class ConfigDialog(QDialog):
text_edit.setMinimumSize(500, 80) # Width, Height text_edit.setMinimumSize(500, 80) # Width, Height
text_edit.setLineWrapMode(QTextEdit.WidgetWidth) text_edit.setLineWrapMode(QTextEdit.WidgetWidth)
self.chunk_size = QSpinBox() self.chunk_size = QSpinBox()
self.chunk_size.setRange(1000, 32000) self.chunk_size.setRange(2000, 128000)
self.chunk_size.setValue(config["CHUNK_SIZE"]) self.chunk_size.setValue(config["CHUNK_SIZE"])
self.max_tokens = QSpinBox() self.max_tokens = QSpinBox()
self.max_tokens.setRange(512, 4096) self.max_tokens.setRange(512, 8000)
self.max_tokens.setValue(config["MAX_TOKENS"]) self.max_tokens.setValue(config["MAX_TOKENS"])
layout.addRow("Whisper URL:", self.whisper_url) layout.addRow("Whisper URL:", self.whisper_url)
@ -342,12 +343,12 @@ class ConfigDialog(QDialog):
layout.addRow("Summary Prompt:", self.summary_prompt) layout.addRow("Summary Prompt:", self.summary_prompt)
layout.addRow("Fact Prompt:", self.fact_prompt) layout.addRow("Fact Prompt:", self.fact_prompt)
layout.addRow("Sentiment Prompt:", self.sentiment_prompt) layout.addRow("Sentiment Prompt:", self.sentiment_prompt)
layout.addRow("Chunk Size:", self.chunk_size) layout.addRow("Tokens Per Chunk:", self.chunk_size)
self.output_dir = QTextEdit(config["OUTPUT_DIR"]) self.output_dir = QTextEdit(config["OUTPUT_DIR"])
self.output_dir.setMinimumSize(500, 80) self.output_dir.setMinimumSize(500, 80)
self.output_dir.setLineWrapMode(QTextEdit.WidgetWidth) self.output_dir.setLineWrapMode(QTextEdit.WidgetWidth)
layout.addRow("Max Tokens:", self.max_tokens) layout.addRow("Max Output Tokens:", self.max_tokens)
layout.addRow("Output Directory:", self.output_dir) layout.addRow("Output Directory:", self.output_dir)
buttons = QHBoxLayout() buttons = QHBoxLayout()