whisper : add large-v3-turbo (#2440)

.gitignore

@@ -3,6 +3,7 @@
 .cache/
 .coreml/
 .test/
+.venv/
 .vs/
 .vscode/
 .DS_Store

Makefile

@@ -1145,8 +1145,9 @@ samples:
 .PHONY: large-v1
 .PHONY: large-v2
 .PHONY: large-v3
+.PHONY: large-v3-turbo
 
-tiny.en tiny base.en base small.en small medium.en medium large-v1 large-v2 large-v3: main
+tiny.en tiny base.en base small.en small medium.en medium large-v1 large-v2 large-v3 large-v3-turbo: main
 	bash ./models/download-ggml-model.sh $@
 	@echo ""
 	@echo "==============================================="

README.md

@@ -236,6 +236,7 @@ make medium
 make large-v1
 make large-v2
 make large-v3
+make large-v3-turbo
 ```
 
 ## Memory usage

bindings/go/examples/go-model-download/main.go

@@ -24,7 +24,7 @@ const (
 
 var (
 	// The models which will be downloaded, if no model is specified as an argument
-	modelNames = []string{"ggml-tiny.en", "ggml-tiny", "ggml-base.en", "ggml-base", "ggml-small.en", "ggml-small", "ggml-medium.en", "ggml-medium", "ggml-large-v1", "ggml-large-v2", "ggml-large-v3"}
+	modelNames = []string{"ggml-tiny.en", "ggml-tiny", "ggml-base.en", "ggml-base", "ggml-small.en", "ggml-small", "ggml-medium.en", "ggml-medium", "ggml-large-v1", "ggml-large-v2", "ggml-large-v3", "large-v3-turbo"}
 )
 
 var (

examples/livestream.sh

@@ -48,7 +48,7 @@ if [ -n "$3" ]; then
 fi
 
 # Whisper models
-models=( "tiny.en" "tiny" "base.en" "base" "small.en" "small" "medium.en" "medium" "large-v1" "large-v2" "large-v3" )
+models=( "tiny.en" "tiny" "base.en" "base" "small.en" "small" "medium.en" "medium" "large-v1" "large-v2" "large-v3" "large-v3-turbo" )
 
 # list available models
 function list_models {

examples/twitch.sh

@@ -21,7 +21,7 @@ help()
     echo "Usage: ./twitch.sh -s [step] -m [model] -t [threads] [url]"
     echo "options:"
     echo "-s       Step in seconds (default is $step)."
-    echo "-m       Choose model, options are: 'tiny.en' 'tiny' 'base.en' 'base' 'small.en' 'small' 'medium.en' 'medium' 'large-v1' 'large-v2' 'large-v3' (default is '$model')."
+    echo "-m       Choose model, options are: 'tiny.en' 'tiny' 'base.en' 'base' 'small.en' 'small' 'medium.en' 'medium' 'large-v1' 'large-v2' 'large-v3' 'large-v3-turbo' (default is '$model')."
     echo "-t       Number of threads to use."
     echo "-h       Print this help page."
     echo

models/README.md

@@ -42,22 +42,24 @@ rmdir models/whisper-medium
 
 ## Available models
 
-| Model         | Disk    | SHA                                        |
-| ------------- | ------- | ------------------------------------------ |
-| tiny          | 75 MiB  | `bd577a113a864445d4c299885e0cb97d4ba92b5f` |
-| tiny.en       | 75 MiB  | `c78c86eb1a8faa21b369bcd33207cc90d64ae9df` |
-| base          | 142 MiB | `465707469ff3a37a2b9b8d8f89f2f99de7299dac` |
-| base.en       | 142 MiB | `137c40403d78fd54d454da0f9bd998f78703390c` |
-| small         | 466 MiB | `55356645c2b361a969dfd0ef2c5a50d530afd8d5` |
-| small.en      | 466 MiB | `db8a495a91d927739e50b3fc1cc4c6b8f6c2d022` |
-| small.en-tdrz | 465 MiB | `b6c6e7e89af1a35c08e6de56b66ca6a02a2fdfa1` |
-| medium        | 1.5 GiB | `fd9727b6e1217c2f614f9b698455c4ffd82463b4` |
-| medium.en     | 1.5 GiB | `8c30f0e44ce9560643ebd10bbe50cd20eafd3723` |
-| large-v1      | 2.9 GiB | `b1caaf735c4cc1429223d5a74f0f4d0b9b59a299` |
-| large-v2      | 2.9 GiB | `0f4c8e34f21cf1a914c59d8b3ce882345ad349d6` |
-| large-v2-q5_0 | 1.1 GiB | `00e39f2196344e901b3a2bd5814807a769bd1630` |
-| large-v3      | 2.9 GiB | `ad82bf6a9043ceed055076d0fd39f5f186ff8062` |
-| large-v3-q5_0 | 1.1 GiB | `e6e2ed78495d403bef4b7cff42ef4aaadcfea8de` |
+| Model               | Disk    | SHA                                        |
+| ------------------- | ------- | ------------------------------------------ |
+| tiny                | 75 MiB  | `bd577a113a864445d4c299885e0cb97d4ba92b5f` |
+| tiny.en             | 75 MiB  | `c78c86eb1a8faa21b369bcd33207cc90d64ae9df` |
+| base                | 142 MiB | `465707469ff3a37a2b9b8d8f89f2f99de7299dac` |
+| base.en             | 142 MiB | `137c40403d78fd54d454da0f9bd998f78703390c` |
+| small               | 466 MiB | `55356645c2b361a969dfd0ef2c5a50d530afd8d5` |
+| small.en            | 466 MiB | `db8a495a91d927739e50b3fc1cc4c6b8f6c2d022` |
+| small.en-tdrz       | 465 MiB | `b6c6e7e89af1a35c08e6de56b66ca6a02a2fdfa1` |
+| medium              | 1.5 GiB | `fd9727b6e1217c2f614f9b698455c4ffd82463b4` |
+| medium.en           | 1.5 GiB | `8c30f0e44ce9560643ebd10bbe50cd20eafd3723` |
+| large-v1            | 2.9 GiB | `b1caaf735c4cc1429223d5a74f0f4d0b9b59a299` |
+| large-v2            | 2.9 GiB | `0f4c8e34f21cf1a914c59d8b3ce882345ad349d6` |
+| large-v2-q5_0       | 1.1 GiB | `00e39f2196344e901b3a2bd5814807a769bd1630` |
+| large-v3            | 2.9 GiB | `ad82bf6a9043ceed055076d0fd39f5f186ff8062` |
+| large-v3-q5_0       | 1.1 GiB | `e6e2ed78495d403bef4b7cff42ef4aaadcfea8de` |
+| large-v3-turbo      | 1.5 GiB | `4af2b29d7ec73d781377bfd1758ca957a807e941` |
+| large-v3-turbo-q5_0 | 547 MiB | `e050f7970618a659205450ad97eb95a18d69c9ee` |
 
 Models are multilingual unless the model name includes `.en`. Models ending in `-q5_0` are [quantized](../README.md#quantization). Models ending in `-tdrz` support local diarization (marking of speaker turns) using [tinydiarize](https://github.com/akashmjn/tinydiarize). More information about models is available [upstream (openai/whisper)](https://github.com/openai/whisper#available-models-and-languages). The list above is a subset of the models supported by the [download-ggml-model.sh](download-ggml-model.sh) script, but many more are available at https://huggingface.co/ggerganov/whisper.cpp/tree/main and elsewhere.
 

models/convert-h5-to-coreml.py

@@ -78,14 +78,14 @@ def convert_hf_whisper(hf_model_name_or_path: str, whisper_state_path: str):
 # Ported from models/convert-whisper-to-coreml.py
 if __name__ == "__main__":
     parser = argparse.ArgumentParser()
-    parser.add_argument("--model-name", type=str, help="name of model to convert (e.g. tiny, tiny.en, base, base.en, small, small.en, medium, medium.en, large-v1, large-v2, large-v3)", required=True)
+    parser.add_argument("--model-name", type=str, help="name of model to convert (e.g. tiny, tiny.en, base, base.en, small, small.en, medium, medium.en, large-v1, large-v2, large-v3, large-v3-turbo)", required=True)
     parser.add_argument("--model-path", type=str, help="path to the model (e.g. if published on HuggingFace: Oblivion208/whisper-tiny-cantonese)", required=True)
     parser.add_argument("--encoder-only", type=bool, help="only convert encoder", default=False)
     parser.add_argument("--quantize",     type=bool, help="quantize weights to F16", default=False)
     parser.add_argument("--optimize-ane", type=bool, help="optimize for ANE execution (currently broken)", default=False)
     args = parser.parse_args()
 
-    if args.model_name not in ["tiny", "tiny.en", "base", "base.en", "small", "small.en", "medium", "medium.en", "large-v1", "large-v2", "large-v3"]:
+    if args.model_name not in ["tiny", "tiny.en", "base", "base.en", "small", "small.en", "medium", "medium.en", "large-v1", "large-v2", "large-v3", "large-v3-turbo"]:
         raise ValueError("Invalid model name")
 
     pt_target_path = f"models/hf-{args.model_name}.pt"

models/convert-whisper-to-coreml.py

@@ -283,13 +283,13 @@ def convert_decoder(hparams, model, quantize=False):
 
 if __name__ == "__main__":
     parser = argparse.ArgumentParser()
-    parser.add_argument("--model", type=str, help="model to convert (e.g. tiny, tiny.en, base, base.en, small, small.en, medium, medium.en, large-v1, large-v2, large-v3)", required=True)
+    parser.add_argument("--model", type=str, help="model to convert (e.g. tiny, tiny.en, base, base.en, small, small.en, medium, medium.en, large-v1, large-v2, large-v3, large-v3-turbo)", required=True)
     parser.add_argument("--encoder-only", type=bool, help="only convert encoder", default=False)
     parser.add_argument("--quantize",     type=bool, help="quantize weights to F16", default=False)
     parser.add_argument("--optimize-ane", type=bool, help="optimize for ANE execution (currently broken)", default=False)
     args = parser.parse_args()
 
-    if args.model not in ["tiny", "tiny.en", "base", "base.en", "small", "small.en", "small.en-tdrz", "medium", "medium.en", "large-v1", "large-v2", "large-v3"]:
+    if args.model not in ["tiny", "tiny.en", "base", "base.en", "small", "small.en", "small.en-tdrz", "medium", "medium.en", "large-v1", "large-v2", "large-v3", "large-v3-turbo"]:
         raise ValueError("Invalid model name")
 
     whisper = load_model(args.model).cpu()

models/convert-whisper-to-openvino.py

@@ -45,10 +45,10 @@ def convert_encoder(hparams, encoder, mname):
 
 if __name__ == "__main__":
     parser = argparse.ArgumentParser()
-    parser.add_argument("--model", type=str, help="model to convert (e.g. tiny, tiny.en, base, base.en, small, small.en, medium, medium.en, large-v1, large-v2, large-v3)", required=True)
+    parser.add_argument("--model", type=str, help="model to convert (e.g. tiny, tiny.en, base, base.en, small, small.en, medium, medium.en, large-v1, large-v2, large-v3, large-v3-turbo)", required=True)
     args = parser.parse_args()
 
-    if args.model not in ["tiny", "tiny.en", "base", "base.en", "small", "small.en", "medium", "medium.en", "large-v1", "large-v2", "large-v3"]:
+    if args.model not in ["tiny", "tiny.en", "base", "base.en", "small", "small.en", "medium", "medium.en", "large-v1", "large-v2", "large-v3", "large-v3-turbo"]:
         raise ValueError("Invalid model name")
 
     whisper = load_model(args.model).cpu()

models/download-coreml-model.sh

@@ -22,7 +22,7 @@ get_script_path() {
 models_path="$(get_script_path)"
 
 # Whisper models
-models="tiny.en tiny base.en base small.en small medium.en medium large-v1 large-v2 large-v3"
+models="tiny.en tiny base.en base small.en small medium.en medium large-v1 large-v2 large-v3 large-v3-turbo"
 
 # list available models
 list_models() {

models/download-ggml-model.cmd

@@ -8,7 +8,7 @@ popd
 set argc=0
 for %%x in (%*) do set /A argc+=1
 
-set models=tiny.en tiny base.en base small.en small medium.en medium large-v1 large-v2 large-v3
+set models=tiny.en tiny base.en base small.en small medium.en medium large-v1 large-v2 large-v3 large-v3-turbo
 
 if %argc% neq 1 (
   echo.

models/download-ggml-model.sh

@@ -46,7 +46,9 @@ large-v1
 large-v2
 large-v2-q5_0
 large-v3
-large-v3-q5_0"
+large-v3-q5_0
+large-v3-turbo
+large-v3-turbo-q5_0"
 
 # list available models
 list_models() {

scripts/bench.py

@@ -64,6 +64,7 @@ models = [
     "ggml-large-v1.bin",
     "ggml-large-v2.bin",
     "ggml-large-v3.bin",
+    "ggml-large-v3-turbo.bin",
 ]
 
 

scripts/convert-all.sh

@@ -1,6 +1,6 @@
 #!/bin/bash
 
-models=( "tiny.en" "tiny" "base.en" "base" "small.en" "small" "medium.en" "medium" "large-v1" "large-v2" "large-v3" )
+models=( "tiny.en" "tiny" "base.en" "base" "small.en" "small" "medium.en" "medium" "large-v1" "large-v2" "large-v3" "large-v3-turbo" )
 
 for model in "${models[@]}"; do
     python3 models/convert-pt-to-ggml.py ~/.cache/whisper/$model.pt ../whisper models/

tests/run-tests.sh

@@ -19,7 +19,7 @@
 cd `dirname $0`
 
 # Whisper models
-models=( "tiny.en" "tiny" "base.en" "base" "small.en" "small" "medium.en" "medium" "large-v1" "large-v2" "large-v3" )
+models=( "tiny.en" "tiny" "base.en" "base" "small.en" "small" "medium.en" "medium" "large-v1" "large-v2" "large-v3" "large-v3-turbo" )
 
 # list available models
 function list_models {