Merge branch 'ggerganov:master' into wyoming

Author: pepijndevos

.gitignore

@@ -58,3 +58,5 @@ cmake-build-debug/
 .cxx/
 .gradle/
 local.properties
+.log
+.exe

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

@@ -9,22 +9,23 @@ import (
 // ContextForSignal returns a context object which is cancelled when a signal
 // is received. It returns nil if no signal parameter is provided
 func ContextForSignal(signals ...os.Signal) context.Context {
-	if len(signals) == 0 {
-		return nil
-	}
+    if len(signals) == 0 {
+        return nil
+    }
 
-	ch := make(chan os.Signal)
-	ctx, cancel := context.WithCancel(context.Background())
+    ch := make(chan os.Signal, 1) // Buffered channel with space for 1 signal
+    ctx, cancel := context.WithCancel(context.Background())
 
-	// Send message on channel when signal received
-	signal.Notify(ch, signals...)
+    // Send message on channel when signal received
+    signal.Notify(ch, signals...)
 
-	// When any signal received, call cancel
-	go func() {
-		<-ch
-		cancel()
-	}()
+    // When any signal is received, call cancel
+    go func() {
+        <-ch
+        cancel()
+    }()
 
-	// Return success
-	return ctx
+    // Return success
+    return ctx
 }
+

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

@@ -9,6 +9,7 @@ import (
 	"net/url"
 	"os"
 	"path/filepath"
+	"strings"
 	"syscall"
 	"time"
 )
@@ -17,14 +18,27 @@ import (
 // CONSTANTS
 
 const (
-	srcUrl  = "https://huggingface.co/ggerganov/whisper.cpp/resolve/main" // The location of the models
-	srcExt  = ".bin"                                                      // Filename extension
-	bufSize = 1024 * 64                                                   // Size of the buffer used for downloading the model
+	srcUrl  = "https://huggingface.co/ggerganov/whisper.cpp/resolve/main/" // The location of the models
+	srcExt  = ".bin"                                                       // Filename extension
+	bufSize = 1024 * 64                                                    // Size of the buffer used for downloading the model
 )
 
 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", "large-v3-turbo"}
+	modelNames = []string{
+		"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",
+	}
 )
 
 var (
@@ -44,7 +58,25 @@ var (
 func main() {
 	flag.Usage = func() {
 		name := filepath.Base(flag.CommandLine.Name())
-		fmt.Fprintf(flag.CommandLine.Output(), "Usage: %s [options] <model>\n\n", name)
+		fmt.Fprintf(flag.CommandLine.Output(), `
+			Usage: %s [options] [<model>...]
+
+			Options:
+  			-out string     Specify the output folder where models will be saved.
+                  			Default: Current working directory.
+  			-timeout duration Set the maximum duration for downloading a model.
+            			      Example: 10m, 1h (default: 30m0s).
+  			-quiet           Suppress all output except errors.
+
+			Examples:
+  			1. Download a specific model:
+     			%s -out ./models tiny-q8_0
+
+			  2. Download all models:
+     			%s -out ./models
+
+			`, name, name, name)
+
 		flag.PrintDefaults()
 	}
 	flag.Parse()
@@ -114,23 +146,87 @@ func GetOut() (string, error) {
 // GetModels returns the list of models to download
 func GetModels() []string {
 	if flag.NArg() == 0 {
-		return modelNames
-	} else {
-		return flag.Args()
+		fmt.Println("No model specified.")
+		fmt.Println("Preparing to download all models...")
+
+		// Calculate total download size
+		fmt.Println("Calculating total download size...")
+		totalSize, err := CalculateTotalDownloadSize(modelNames)
+		if err != nil {
+			fmt.Println("Error calculating download sizes:", err)
+			os.Exit(1)
+		}
+
+		fmt.Println("View available models: https://huggingface.co/ggerganov/whisper.cpp/tree/main")
+		fmt.Printf("Total download size: %.2f GB\n", float64(totalSize)/(1024*1024*1024))
+		fmt.Println("Would you like to download all models? (y/N)")
+
+		// Prompt for user input
+		var response string
+		fmt.Scanln(&response)
+		if response != "y" && response != "Y" {
+			fmt.Println("Aborting. Specify a model to download.")
+			os.Exit(0)
+		}
+
+		return modelNames // Return all models if confirmed
 	}
+	return flag.Args() // Return specific models if arguments are provided
+}
+
+func CalculateTotalDownloadSize(models []string) (int64, error) {
+	var totalSize int64
+	client := http.Client{}
+
+	for _, model := range models {
+		modelURL, err := URLForModel(model)
+		if err != nil {
+			return 0, err
+		}
+
+		// Issue a HEAD request to get the file size
+		req, err := http.NewRequest("HEAD", modelURL, nil)
+		if err != nil {
+			return 0, err
+		}
+
+		resp, err := client.Do(req)
+		if err != nil {
+			return 0, err
+		}
+		resp.Body.Close()
+
+		if resp.StatusCode != http.StatusOK {
+			fmt.Printf("Warning: Unable to fetch size for %s (HTTP %d)\n", model, resp.StatusCode)
+			continue
+		}
+
+		size := resp.ContentLength
+		totalSize += size
+	}
+	return totalSize, nil
 }
 
 // URLForModel returns the URL for the given model on huggingface.co
 func URLForModel(model string) (string, error) {
+	// Ensure "ggml-" prefix is added only once
+	if !strings.HasPrefix(model, "ggml-") {
+		model = "ggml-" + model
+	}
+
+	// Ensure ".bin" extension is added only once
 	if filepath.Ext(model) != srcExt {
 		model += srcExt
 	}
+
+	// Parse the base URL
 	url, err := url.Parse(srcUrl)
 	if err != nil {
 		return "", err
-	} else {
-		url.Path = filepath.Join(url.Path, model)
 	}
+
+	// Ensure no trailing slash in the base URL
+	url.Path = fmt.Sprintf("%s/%s", strings.TrimSuffix(url.Path, "/"), model)
 	return url.String(), nil
 }
 

examples/addon.node/addon.cpp

@@ -1,5 +1,6 @@
 #include "napi.h"
 #include "common.h"
+#include "common-whisper.h"
 
 #include "whisper.h"
 

examples/common-sdl.cpp

@@ -159,15 +159,11 @@ void audio_async::callback(uint8_t * stream, int len) {
 
             memcpy(&m_audio[m_audio_pos], stream, n0 * sizeof(float));
             memcpy(&m_audio[0], stream + n0 * sizeof(float), (n_samples - n0) * sizeof(float));
-
-            m_audio_pos = (m_audio_pos + n_samples) % m_audio.size();
-            m_audio_len = m_audio.size();
         } else {
             memcpy(&m_audio[m_audio_pos], stream, n_samples * sizeof(float));
-
-            m_audio_pos = (m_audio_pos + n_samples) % m_audio.size();
-            m_audio_len = std::min(m_audio_len + n_samples, m_audio.size());
         }
+        m_audio_pos = (m_audio_pos + n_samples) % m_audio.size();
+        m_audio_len = std::min(m_audio_len + n_samples, m_audio.size());
     }
 }
 

examples/common-whisper.cpp

@@ -76,30 +76,25 @@ bool read_audio_data(const std::string & fname, std::vector<float>& pcmf32, std:
 
 		fprintf(stderr, "%s: read %zu bytes from stdin\n", __func__, audio_data.size());
     }
-    else if (is_wav_buffer(fname)) {
-			if ((result = ma_decoder_init_memory(audio_data.data(), audio_data.size(), &decoder_config, &decoder)) != MA_SUCCESS) {
-				fprintf(stderr, "Error: failed to open audio data from fname buffer (%s)\n", ma_result_description(result));
-
-				return false;
-			}
-    }
-    else if ((result = ma_decoder_init_file(fname.c_str(), &decoder_config, &decoder)) != MA_SUCCESS) {
+    else if (((result = ma_decoder_init_file(fname.c_str(), &decoder_config, &decoder)) != MA_SUCCESS)) {
 #if defined(WHISPER_FFMPEG)
-			if (ffmpeg_decode_audio(fname, audio_data) != 0) {
-				fprintf(stderr, "error: failed to ffmpeg decode '%s'\n", fname.c_str());
+		if (ffmpeg_decode_audio(fname, audio_data) != 0) {
+			fprintf(stderr, "error: failed to ffmpeg decode '%s'\n", fname.c_str());
 
-				return false;
-			}
+			return false;
+		}
 
-			if ((result = ma_decoder_init_memory(audio_data.data(), audio_data.size(), &decoder_config, &decoder)) != MA_SUCCESS) {
-				fprintf(stderr, "error: failed to read audio data as wav (%s)\n", ma_result_description(result));
+		if ((result = ma_decoder_init_memory(audio_data.data(), audio_data.size(), &decoder_config, &decoder)) != MA_SUCCESS) {
+			fprintf(stderr, "error: failed to read audio data as wav (%s)\n", ma_result_description(result));
 
-				return false;
-			}
+			return false;
+		}
 #else
-		fprintf(stderr, "error: failed to open '%s' file (%s)\n", fname.c_str(), ma_result_description(result));
+		if ((result = ma_decoder_init_memory(fname.c_str(), fname.size(), &decoder_config, &decoder)) != MA_SUCCESS) {
+			fprintf(stderr, "error: failed to read audio data as wav (%s)\n", ma_result_description(result));
 
-		return false;
+			return false;
+		}
 #endif
     }
 

examples/common.cpp

@@ -609,21 +609,6 @@ gpt_vocab::id gpt_sample_top_k_top_p_repeat(
 
 }
 
-bool is_wav_buffer(const std::string buf) {
-    // RIFF ref: https://en.wikipedia.org/wiki/Resource_Interchange_File_Format
-    // WAV ref: https://www.mmsp.ece.mcgill.ca/Documents/AudioFormats/WAVE/WAVE.html
-    if (buf.size() < 12 || buf.substr(0, 4) != "RIFF" || buf.substr(8, 4) != "WAVE") {
-        return false;
-    }
-
-    uint32_t chunk_size = *reinterpret_cast<const uint32_t*>(buf.data() + 4);
-    if (chunk_size + 8 != buf.size()) {
-        return false;
-    }
-
-    return true;
-}
-
 void high_pass_filter(std::vector<float> & data, float cutoff, float sample_rate) {
     const float rc = 1.0f / (2.0f * M_PI * cutoff);
     const float dt = 1.0f / sample_rate;

examples/common.h

@@ -134,9 +134,6 @@ gpt_vocab::id gpt_sample_top_k_top_p_repeat(
 // Audio utils
 //
 
-// Check if a buffer is a WAV audio file
-bool is_wav_buffer(const std::string buf);
-
 // Write PCM data into WAV audio file
 class wav_writer {
 private:

models/download-ggml-model.cmd

@@ -8,7 +8,18 @@ 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 large-v3-turbo
+set models=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
 
 if %argc% neq 1 (
   echo.
@@ -50,7 +61,7 @@ if %ERRORLEVEL% neq 0 (
 
 echo Done! Model %model% saved in %root_path%\models\ggml-%model%.bin
 echo You can now use it like this:
-echo build\bin\Release\whisper-cli.exe -m %root_path%\models\ggml-%model%.bin -f %root_path%\samples\jfk.wav
+echo %~dp0build\bin\Release\whisper-cli.exe -m %root_path%\models\ggml-%model%.bin -f %root_path%\samples\jfk.wav
 
 goto :eof