From 7d2498e9c94aa13144d0dc05adbfa2f3ba81f252 Mon Sep 17 00:00:00 2001 From: Alex Blank Date: Wed, 1 Jan 2025 16:24:07 +0100 Subject: [PATCH] more fixes --- config.yml | 6 +++--- src/annescribe/transcription/job_utils.py | 6 +++++- 2 files changed, 8 insertions(+), 4 deletions(-) diff --git a/config.yml b/config.yml index f7bfcba..6008ce7 100644 --- a/config.yml +++ b/config.yml @@ -6,9 +6,9 @@ data: audio: chunking: - min_silence_level: -40 - min_silence_length: 200 - ms_silence_to_keep: 100 + min_silence_level: -30 + min_silence_length: 400 + ms_silence_to_keep: 50 model: type: large diff --git a/src/annescribe/transcription/job_utils.py b/src/annescribe/transcription/job_utils.py index db682ab..6460159 100644 --- a/src/annescribe/transcription/job_utils.py +++ b/src/annescribe/transcription/job_utils.py @@ -246,11 +246,15 @@ def get_progress_details(job_name: str, config: dict) -> dict: } newest_output_file = max([os.path.getctime(os.path.join(output_dir, f)) for f in os.listdir(output_dir)]) + oldest_output_file = min([os.path.getctime(os.path.join(output_dir, f)) for f in os.listdir(output_dir)]) + first_chunk_time = datetime.fromtimestamp(oldest_output_file) latest_chunk_time = datetime.fromtimestamp(newest_output_file) # calculate time remaining - time_remaining = (latest_chunk_time - start_time) / num_output_files * (num_input_files - num_output_files) + time_elapsed = latest_chunk_time - first_chunk_time + time_per_chunk = time_elapsed / num_output_files + time_remaining = time_per_chunk * (num_input_files - num_output_files) # format time remaining time_remaining_formatted = ""