Fine-tune transcription

Bump up recording window to 28 seconds. This helps a lot with long-form
transcription tasks, s.a. transcribing an audiobook.

We should expose this as a parameter, since at 10s the transcription delay is
typically 300ms, while at 28s it's typically 1.1-1.2s.

1 files changed, 2 insertions, 2 deletions

diff --git a/Scripts/transcribe.py b/Scripts/transcribe.py
index 37a8407..21bb4ba 100644
--- a/Scripts/transcribe.py
+++ b/Scripts/transcribe.py
@@ -35,7 +35,7 @@ class AudioState:
 
         # The maximum length that recordAudio() will put into frames before it
         # starts dropping from the start.
-        self.MAX_LENGTH_S = 10
+        self.MAX_LENGTH_S = 28
         self.MAX_LENGTH_S_WHISPER = 30
         # The minimum length that recordAudio() will wait for before saving audio.
         self.MIN_LENGTH_S = 1
@@ -283,7 +283,7 @@ def transcribeAudio(audio_state, model):
         old_text = audio_state.text
 
         audio_state.text = string_matcher.matchStrings(audio_state.text,
-                text, window_size = 20)
+                text, window_size = 25)
         if old_text != audio_state.text:
             # We think the user said something, so  reset the amount of
             # time we sleep between transcriptions to the minimum.