go-whisper

Docker Image for Speech-to-Text using ggerganov/whisper.cpp.

This Docker image provides a ready-to-use environment for converting speech to text using the ggerganov/whisper.cpp library. The whisper.cpp library is an open-source project that enables efficient and accurate speech recognition. By utilizing this Docker image, users can easily set up and run the speech-to-text conversion process without worrying about installing dependencies or configuring the system.

The Docker image includes all necessary components and dependencies, ensuring a seamless experience for users who want to leverage the power of the whisper.cpp library for their speech recognition needs. Simply pull the Docker image, run the container, and start converting your audio files into text with minimal effort.

In summary, this Docker image offers a convenient and efficient way to utilize the ggerganov/whisper.cpp library for speech-to-text conversion, making it an excellent choice for those looking to implement speech recognition in their projects.

OpenAI's Whisper models converted to ggml format

See the Available models.

For more information see ggerganov/whisper.cpp.

Prepare

Download the model you want to use and put it in the models directory.

curl -LJ https://huggingface.co/ggerganov/whisper.cpp/resolve/main/ggml-small.bin \
  --output models/ggml-small.bin

Usage

Please follow these simplified instructions to transcribe the audio file using a Docker container:

1. Ensure that you have a testdata directory containing the jfk.wav file.
2. Mount both the models and testdata directories to the Docker container.
3. Specify the model using the --model flag and the audio file path using the --audio-path flag.
4. The transcript result file will be saved in the same directory as the audio file.

To transcribe the audio file, execute the command provided below.

docker run \
  -v $PWD/models:/app/models \
  -v $PWD/testdata:/app/testdata \
  ghcr.io/appleboy/go-whisper:latest \
  --model /app/models/ggml-small.bin \
  --audio-path /app/testdata/jfk.wav

See the following output:

whisper_init_from_file_no_state: loading model from '/app/models/ggml-small.bin'
whisper_model_load: loading model
whisper_model_load: n_vocab       = 51865
whisper_model_load: n_audio_ctx   = 1500
whisper_model_load: n_audio_state = 768
whisper_model_load: n_audio_head  = 12
whisper_model_load: n_audio_layer = 12
whisper_model_load: n_text_ctx    = 448
whisper_model_load: n_text_state  = 768
whisper_model_load: n_text_head   = 12
whisper_model_load: n_text_layer  = 12
whisper_model_load: n_mels        = 80
whisper_model_load: ftype         = 1
whisper_model_load: qntvr         = 0
whisper_model_load: type          = 3
whisper_model_load: mem required  =  743.00 MB (+   16.00 MB per decoder)
whisper_model_load: adding 1608 extra tokens
whisper_model_load: model ctx     =  464.68 MB
whisper_model_load: model size    =  464.44 MB
whisper_init_state: kv self size  =   15.75 MB
whisper_init_state: kv cross size =   52.73 MB
1:46AM INF system_info: n_threads = 8 / 8 | AVX = 0 | AVX2 = 0 | AVX512 = 0 | FMA = 0 | NEON = 1 | ARM_FMA = 1 | F16C = 0 | FP16_VA = 0 | WASM_SIMD = 0 | BLAS = 0 | SSE3 = 0 | VSX = 0 | COREML = 0 | 
 module=transcript
whisper_full_with_state: auto-detected language: en (p = 0.967331)
1:46AM INF [    0s ->    11s] And so my fellow Americans, ask not what your country can do for you, ask what you can do for your country. module=transcript