add new models

Author: jmgirard

DESCRIPTION

@@ -29,6 +29,5 @@ Suggests:
     audio.vadwebrtc (>= 0.2.0)
 LinkingTo: Rcpp
 SystemRequirements: GNU make
-RoxygenNote: 7.1.2
+RoxygenNote: 7.3.2
 Remotes: bnosac/audio.vadwebrtc
-

R/whisper.R

@@ -237,12 +237,18 @@ align_skipped <- function(sentences, skipped, from = "from", to = "to"){
 #'                  language = "en", duration = 1000)
 #' }
 whisper <- function(x, use_gpu = FALSE, overwrite = FALSE, model_dir = Sys.getenv("WHISPER_MODEL_DIR", unset = getwd()), ...){
-  if(x %in% c("tiny", "tiny.en", "base", "base.en", "small", "small.en", "medium", "medium.en", "large-v1", "large-v2", "large-v3", "large",
-              "tiny-q5_1", "tiny.en-q5_1", 
-              "base-q5_1", "base.en-q5_1", 
-              "small-q5_1", "small.en-q5_1", 
-              "medium-q5_0", "medium.en-q5_0", 
-              "large-v2-q5_0", "large-v3-q5_0")){
+  if(x %in% c("tiny", "tiny-q5_1", "tiny-q8_0",
+              "tiny.en", "tiny.en-q5_1", "tiny.en-q8_0",
+              "base", "base-q5_1", "base-q8_0", 
+              "base.en", "base.en-q5_1", "base.en-q8_0",
+              "small", "small-q5_1", "small-q8_0",
+              "small.en", "small.en-q5_1", "small.en-q8_0",
+              "medium", "medium-q5_0", "medium-q8_0",
+              "medium.en", "medium.en-q5_0", "medium.en-q8_0", 
+              "large-v1", 
+              "large-v2", "large-v2-q5_0", "large-v2-q8_0", 
+              "large-v3", "large-v3-q5_0",
+              "large-v3-turbo", "large-v3-turbo-q5_0", "large-v3-turbo-q8_0")){
     x <- whisper_download_model(x, overwrite = overwrite, model_dir = model_dir)
   }
   if(inherits(x, "whisper_download")){
@@ -263,8 +269,8 @@ whisper <- function(x, use_gpu = FALSE, overwrite = FALSE, model_dir = Sys.geten
 #' \item{base & base.en: 142 MB, RAM required: ~500 MB. Multilingual and English only version.}
 #' \item{small & small.en: 466 MB, RAM required: ~1.0 GB. Multilingual and English only version.}
 #' \item{medium & medium.en: 1.5 GB, RAM required: ~2.6 GB. Multilingual and English only version.}
-#' \item{large-v1, large-v2, large-v3: 2.9 GB, RAM required: ~4.7 GB. Multilingual}
-#' \item{quantised models: tiny-q5_1, tiny.en-q5_1, base-q5_1, base.en-q5_1, small-q5_1, small.en-q5_1, medium-q5_0, medium.en-q5_0, large-v2-q5_0 and large-v3-q5_0 (only - from version 1.5.4 onwards)}
+#' \item{large-v1, large-v2, large-v3, large-v3-turbo: 2.9 GB, RAM required: ~4.7 GB. Multilingual}
+#' \item{quantised models: tiny-q5_1, tiny-q8_0, tiny.en-q5_1, tiny.en-q8_0, base-q5_1, base-q8_0, base.en-q5_1, base_q8_0, small-q5_1, small-q8_0, small.en-q5_1, small.en-q8_0, medium-q5_0, medium-q8_0, medium.en-q5_0, medium.en-q8_0, large-v2-q5_0, large-v2-q8_0, large-v3-q5_0, large-v3-turbo-q5_0, and large-v3-turbo-q8_0 (only - from version 1.5.4 onwards)}
 #' }
 #' Note that the larger models may take longer than 60 seconds to download, so consider 
 #' increasing the timeout option in R via \code{options(timeout = 120)}
@@ -314,12 +320,18 @@ whisper <- function(x, use_gpu = FALSE, overwrite = FALSE, model_dir = Sys.geten
 #' \dontshow{
 #' if(file.exists(path$file_model)) file.remove(path$file_model)
 #' }
-whisper_download_model <- function(x = c("tiny", "tiny.en", "base", "base.en", "small", "small.en", "medium", "medium.en", "large-v1", "large-v2", "large-v3", "large",
-                                         "tiny-q5_1", "tiny.en-q5_1", 
-                                         "base-q5_1", "base.en-q5_1", 
-                                         "small-q5_1", "small.en-q5_1", 
-                                         "medium-q5_0", "medium.en-q5_0", 
-                                         "large-v2-q5_0", "large-v3-q5_0"),
+whisper_download_model <- function(x = c("tiny", "tiny-q5_1", "tiny-q8_0",
+                                         "tiny.en", "tiny.en-q5_1", "tiny.en-q8_0",
+                                         "base", "base-q5_1", "base-q8_0", 
+                                         "base.en", "base.en-q5_1", "base.en-q8_0",
+                                         "small", "small-q5_1", "small-q8_0",
+                                         "small.en", "small.en-q5_1", "small.en-q8_0",
+                                         "medium", "medium-q5_0", "medium-q8_0",
+                                         "medium.en", "medium.en-q5_0", "medium.en-q8_0", 
+                                         "large-v1", 
+                                         "large-v2", "large-v2-q5_0", "large-v2-q8_0", 
+                                         "large-v3", "large-v3-q5_0",
+                                         "large-v3-turbo", "large-v3-turbo-q5_0", "large-v3-turbo-q8_0"),
                                    model_dir = Sys.getenv("WHISPER_MODEL_DIR", unset = getwd()),
                                    repos = c("huggingface", "ggerganov"),
                                    version = c("1.5.4", "1.2.1"),

man/whisper_download_model.Rd

@@ -5,6 +5,11 @@
 
 using namespace Rcpp;
 
+#ifdef RCPP_USE_GLOBAL_ROSTREAM
+Rcpp::Rostream<true>&  Rcpp::Rcout = Rcpp::Rcpp_cout_get();
+Rcpp::Rostream<false>& Rcpp::Rcerr = Rcpp::Rcpp_cerr_get();
+#endif
+
 // whisper_load_model
 SEXP whisper_load_model(std::string model, bool use_gpu);
 RcppExport SEXP _audio_whisper_whisper_load_model(SEXP modelSEXP, SEXP use_gpuSEXP) {