From 696663549927b814b7f1e9596add37196e7685f6 Mon Sep 17 00:00:00 2001 From: Christopher Oezbek Date: Fri, 4 Jul 2025 13:37:07 +0200 Subject: [PATCH] Support for stereo audio --- scripts/stt_from_file_pytorch.py | 1 + 1 file changed, 1 insertion(+) diff --git a/scripts/stt_from_file_pytorch.py b/scripts/stt_from_file_pytorch.py index fd10c67..d4b4dc0 100644 --- a/scripts/stt_from_file_pytorch.py +++ b/scripts/stt_from_file_pytorch.py @@ -152,6 +152,7 @@ def main(args): audio, input_sample_rate = sphn.read(args.in_file) audio = torch.from_numpy(audio).to(args.device) + audio = audio.mean(dim=0, keepdim=True) # Convert to mono audio = julius.resample_frac(audio, input_sample_rate, mimi.sample_rate) if audio.shape[-1] % mimi.frame_size != 0: to_pad = mimi.frame_size - audio.shape[-1] % mimi.frame_size