Pipeline to Pipeline transcription model built off of the WhisperX library with Pyannote. Comes with Word Alignment, and Speaker Diarization (identifying who said what).
Note: The second branch is a version of the model catering to background noise which is specifically other people talking, but this version can also deal with other types of backgroud noise.
This transcription tool stays entirely local to your device, so your data and your audios are too, whereas in commercial software you pay for, your data/audio is often sent to a 3rd party cloud or database that is out of your control. With this, one can have data sovereignty at no cost.
Features: Processing: Uses a checkpoint system to save progress after each stage (Transcribe → Align → Diarize), so you dont lose too much work if there are issues, like with your device or the environment (you can change the checkpoint value, and batch number).
Workflow: Creates and manages project folders for audio inputs, checkpoints, and final transcripts.
Speaker Mapping: Dictionary to convert eg. "SPEAKER_00" into real names.
I built this on my Mac but it should run fine on other OS's. There is a caffeine function leftover for Mac, as Macs can have issues with the configurations and sometimes take longer, but this may not be necessary, especially a year later when libraries have had updates/improvements.
This project is currently configured to run on CPU. While macOS has MPS (Metal) support, WhisperX currently provides better stability on CPU for the transcription/alignment pipeline.
Mandatory: You must have ffmpeg installed on your system: Brew install ffmpeg
Note: To bypass security blocks introduced in PyTorch 2.6+ that are not compatible with Pyannote models, this project uses a custom trusted_load patch.
Recommended Environment:
Python: 3.10
PyTorch: 2.5.1
How to Run:
Create a Main Folder on your Desktop named TranscriptionProject.
Place your audio file inside TranscriptionProject/audio_input/.
Hugging Face Token: Make sure you have generated a Hugging Face token (this is free) with access to the pyannote/speaker-diarization-3.1 model.
In the code, find and set the min_speakers and max_speakers values to the number of speakers in the audio, this helps force the model to choose between voice "blueprints" that have already been created.
Structure: audio_input/: Place your .m4a or .wav files here.
checkpoints/: Stores checkpoint.json so you can resume if the script crashes.
outputs/: Your final, speaker-labeled transcript.txt will appear here.
Package requirements below -
Package Version
aiohappyeyeballs 2.6.1 aiohttp 3.13.3 aiosignal 1.4.0 alembic 1.18.1 antlr4-python3-runtime 4.9.3 asteroid-filterbanks 0.4.0 async-timeout 5.0.1 attrs 25.4.0 av 16.1.0 certifi 2026.1.4 cffi 2.0.0 charset-normalizer 3.4.4 click 8.3.1 coloredlogs 15.0.1 colorlog 6.10.1 contourpy 1.3.2 ctranslate2 4.6.3 cycler 0.12.1 docopt 0.6.2 einops 0.8.2 faster-whisper 1.2.1 filelock 3.20.3 flatbuffers 25.12.19 fonttools 4.61.1 frozenlist 1.8.0 fsspec 2026.1.0 hf-xet 1.2.0 huggingface-hub 0.36.0 humanfriendly 10.0 HyperPyYAML 1.2.3 idna 3.11 Jinja2 3.1.6 joblib 1.5.3 julius 0.2.7 kiwisolver 1.4.9 lightning 2.6.0 lightning-utilities 0.15.2 Mako 1.3.10 markdown-it-py 4.0.0 MarkupSafe 3.0.3 matplotlib 3.10.8 mdurl 0.1.2 mpmath 1.3.0 multidict 6.7.1 networkx 3.4.2 nltk 3.9.2 numpy 2.2.6 omegaconf 2.3.0 onnxruntime 1.23.2 optuna 4.7.0 packaging 26.0 pandas 2.3.3 pillow 12.1.0 pip 25.0.1 primePy 1.3 propcache 0.4.1 protobuf 6.33.4 pyannote.audio 3.4.0 pyannote.core 5.0.0 pyannote.database 5.1.3 pyannote.metrics 3.2.1 pyannote.pipeline 3.0.1 pycparser 3.0 Pygments 2.19.2 pyparsing 3.3.2 python-dateutil 2.9.0.post0 pytorch-lightning 2.6.0 pytorch-metric-learning 2.9.0 pytz 2025.2 PyYAML 6.0.3 regex 2026.1.15 requests 2.32.5 rich 14.3.1 ruamel.yaml 0.18.17 ruamel.yaml.clib 0.2.15 safetensors 0.7.0 scikit-learn 1.7.2 scipy 1.15.3 semver 3.0.4 sentencepiece 0.2.1 setuptools 78.1.0 shellingham 1.5.4 six 1.17.0 sortedcontainers 2.4.0 soundfile 0.13.1 speechbrain 1.0.3 SQLAlchemy 2.0.46 sympy 1.13.1 tabulate 0.9.0 tensorboardX 2.6.4 threadpoolctl 3.6.0 tokenizers 0.22.2 tomli 2.4.0 torch 2.5.1 torch-audiomentations 0.12.0 torch_pitch_shift 1.2.5 torchaudio 2.5.1 torchmetrics 1.8.2 tqdm 4.67.1 transformers 4.57.6 typer 0.21.1 typing_extensions 4.15.0 tzdata 2025.3 urllib3 2.6.3 wheel 0.45.1 whisperx 3.7.6 yarl 1.22.0