Merge pull request #16 from langtech-bsc/feature/box-translation

whisperlivekit/whisper_streaming_custom/backends.py: separator should be an empty space

whisperlivekit/audio_processor.py: bugfix when lines array is empty

Dockerfile: use latest BSC model BSC-LT/faster-whisper-large-v3-ca-punctuated-3370h

Merge pull request #15 from langtech-bsc/feature/box-translation

resolve merge in Dockerfile

Dockerfile: remove commented line

Dockerfile: ASR with large-v2

Merge pull request #14 from langtech-bsc/feature/box-translation

Dockerfile: switch to CA-EN in MT task

scripts/launch_server.sh: large-v2 for ASR, CA-EN for MR

scripts/launch_server.sh: MT test with CA-EN model

Merge pull request #13 from langtech-bsc/feature/box-translation

Merge pull request #12 from langtech-bsc/feature/box-translation

whisperlivekit/audio_processor.py: apply translation to the whole recognized chunk insted of on a token basis

Merge pull request #11 from langtech-bsc/feature/box-translation

Dockerfile: test to record demo

Merge pull request #10 from langtech-bsc/feature/box-translation

Dockerfile: solve merge

Dockerfile: using model with punctuation

Dockerfile: test with ca 2 es with large-v3

Dockerfile: test with es 2 ca

Revert "whisperlivekit/audio_processor.py: check with only ASR, not MT"

whisperlivekit/audio_processor.py: check with only ASR, not MT

Dockerfile: set buffer_trimming_sec to 2 seconds when launching the server

Merge pull request #9 from langtech-bsc/feature/box-translation

whisperlivekit/audio_processor.py: use space as separator

Merge pull request #8 from langtech-bsc/feature/box-translation

whisperlivekit/audio_processor.py: translate only the new text coming from transcription

Merge pull request #7 from langtech-bsc/feature/box-translation

whisperlivekit/core.py: test ctranslate2.Translator with device = cuda

Merge pull request #6 from langtech-bsc/feature/box-translation

whisperlivekit/core.py: ensure ctranslate2.Translator is loaded in the cpu

whisperlivekit/audio_processor.py: activate translation

Merge pull request #5 from langtech-bsc/feature/box-translation

Merge pull request #4 from langtech-bsc/feature/box-translation

whisperlivekit/whisper_streaming_custom/backends.py: use BatchedInferencePipeline in FasterWhisperASR class, and batch_size=16 in transcribe() function, also created WhisperXASR class

Dockerfile: do ASR with model projecte-aina/faster-whisper-large-v3-ca-3catparla

Merge pull request #3 from langtech-bsc/feature/box-translation

Revert "Dockerfile whisperlivekit/audio_processor.py: change fils to do ASR and MT with aina model"

Dockerfile whisperlivekit/audio_processor.py: change fils to do ASR and MT with aina model

Dockerfile: add usage of flag --model_cascaded_translation

Merge pull request #2 from langtech-bsc/feature/box-translation

Merge pull request #1 from langtech-bsc/feature/box-translation

whisperlivekit/audio_processor.py: translation controlled by and if/else, added await to call translate_text as async

scripts/: some useful bash scripts to launch the server and do port forwarding

connected translation with frontend

whisperlivekit/core.py: print args, load model_cascaded_translation from input parameter

whisperlivekit/audio_processor.py: bug fix to translate from transcribed text and show in log messages

whisperlivekit/core.py: load models for translation in class WhisperLiveKit