Leo Huang

Crash on iPhone when Using CoreML

1

Followed instruction in README to convert coreml model. Then tested on macOS, it works perfectly. Then copy the model to SwiftUI example project, followed by adding WHISPER_USE_COREML processor and coreml...

Failed to generate timestamp for nvidia/parakeet-tdt-1.1b

4

**Describe the bug** When I tried to generate timestamp with model: nvidia/parakeet-tdt-1.1b, I got following error, ValueError: `char_offsets`: [{'char': [tensor(607, dtype=torch.int32)], 'start_offset': 28, 'end_offset': 29}.... call stack, ``` Traceback (most...

Is that possible to convert the model to ONNX then use it in C++

10

Is that possible to convert the model to ONNX then use it in C++ for speaker diarization? Thanks.

Inconsistent result when use different embeddin config

3

I used following code to do speaker diarization, ``` from pyannote.audio import Pipeline pipeline = Pipeline.from_pretrained("pyannote/[email protected]", use_auth_token="ACCESS_TOKEN_GOES_HERE") # apply the pipeline to an audio file diarization = pipeline("audio.wav") ``` After...

Is it possible that export model to something that can be used in C++ using libtorch?

Is possible to generate same result of pytorch stft

1

Thanks for great work. I want to use this library to mimic pytorch STFT, so that given same signal( wav data ), generate same output. I used test.cpp with same...

Duplicate words generated

17

I used latest **commit**: [bf2449d](https://github.com/ggerganov/whisper.cpp/commit/bf2449dfae35a46b2cd92ab22661ce81a48d4993) with model: ggml-small.bin by executing **command**, $> bin/main -m ../models/ggml-small.bin ~/tmp/wrongResultWithWhisper.wav in macOS. Output has many **duplicate words** as below, [00:00:33.000 --> 00:00:44.000] To this...