diff options
| author | yum <yum.food.vr@gmail.com> | 2022-10-16 23:48:15 -0700 |
|---|---|---|
| committer | yum <yum.food.vr@gmail.com> | 2022-10-17 00:24:57 -0700 |
| commit | 247f163efd46a58b2fbb5e7e26e0d141252dc651 (patch) | |
| tree | 97354f82193410360f1606d779b1facfaf1f4f90 /osc_ctrl.py | |
| parent | 6eb03b1e369286a3e0417b3229f50f00f5760e8a (diff) | |
Add continuous transcription mode
Algorithm:
* look at last 20 chars of last committed transcription
* scan new transcription using 10-char sliding window
* find spot where distance is minimized
* stitch two messages together
Thus we're able to maintain a continuously growing transcription
without having to feed the AI more than 30 seconds of data at a
time. Seems to work reasonably well in bench tests.
Also fix silence detection. AI exposes a probability that nothing
was said. Hand-pick a probability of 0.1. Sometimes the AI still
goes sicko mode with this setting but going higher occasionally
results in no transcription.
Diffstat (limited to 'osc_ctrl.py')
| -rw-r--r-- | osc_ctrl.py | 2 |
1 files changed, 0 insertions, 2 deletions
diff --git a/osc_ctrl.py b/osc_ctrl.py index d80f055..761ff78 100644 --- a/osc_ctrl.py +++ b/osc_ctrl.py @@ -241,8 +241,6 @@ def resizeBoard(num_lines, tx_state, shrink_only): # This may take multiple calls to complete. Returns True once it's done. def sendMessageLazy(client, msg, tx_state): lines = splitMessage(msg) - #resizeBoard(len(lines), tx_state, shrink_only=False) - msg_encoded = encodeMessage(lines) msg_encoded_len = len(msg_encoded) |