Do not crash transcription system if the language code is unknown

bcc-code · Aug 7, 2024 · ad166d5 · ad166d5
1 parent 554c479
commit ad166d5
Showing 1 changed file with 119 additions and 0 deletions.
diff --git a/services/transcribe/transcribe.go b/services/transcribe/transcribe.go
@@ -3,6 +3,7 @@ package transcribe
 import (
 	"context"
 	"fmt"
+	"strings"
 	"time"
 
 	"github.com/bcc-code/bcc-media-flows/common"
@@ -70,6 +71,124 @@ func DebugResponse(resp *resty.Response) {
 	fmt.Println("  RemoteAddr    :", ti.RemoteAddr.String())
 }
 
+var whisperSupportedLanguages = map[string]bool{
+	"en":  true,
+	"zh":  true,
+	"de":  true,
+	"es":  true,
+	"ru":  true,
+	"ko":  true,
+	"fr":  true,
+	"ja":  true,
+	"pt":  true,
+	"tr":  true,
+	"pl":  true,
+	"ca":  true,
+	"nl":  true,
+	"ar":  true,
+	"sv":  true,
+	"it":  true,
+	"id":  true,
+	"hi":  true,
+	"fi":  true,
+	"vi":  true,
+	"he":  true,
+	"uk":  true,
+	"el":  true,
+	"ms":  true,
+	"cs":  true,
+	"ro":  true,
+	"da":  true,
+	"hu":  true,
+	"ta":  true,
+	"no":  true,
+	"th":  true,
+	"ur":  true,
+	"hr":  true,
+	"bg":  true,
+	"lt":  true,
+	"la":  true,
+	"mi":  true,
+	"ml":  true,
+	"cy":  true,
+	"sk":  true,
+	"te":  true,
+	"fa":  true,
+	"lv":  true,
+	"bn":  true,
+	"sr":  true,
+	"az":  true,
+	"sl":  true,
+	"kn":  true,
+	"et":  true,
+	"mk":  true,
+	"br":  true,
+	"eu":  true,
+	"is":  true,
+	"hy":  true,
+	"ne":  true,
+	"mn":  true,
+	"bs":  true,
+	"kk":  true,
+	"sq":  true,
+	"sw":  true,
+	"gl":  true,
+	"mr":  true,
+	"pa":  true,
+	"si":  true,
+	"km":  true,
+	"sn":  true,
+	"yo":  true,
+	"so":  true,
+	"af":  true,
+	"oc":  true,
+	"ka":  true,
+	"be":  true,
+	"tg":  true,
+	"sd":  true,
+	"gu":  true,
+	"am":  true,
+	"yi":  true,
+	"lo":  true,
+	"uz":  true,
+	"fo":  true,
+	"ht":  true,
+	"ps":  true,
+	"tk":  true,
+	"nn":  true,
+	"mt":  true,
+	"sa":  true,
+	"lb":  true,
+	"my":  true,
+	"bo":  true,
+	"tl":  true,
+	"mg":  true,
+	"as":  true,
+	"tt":  true,
+	"haw": true,
+	"ln":  true,
+	"ha":  true,
+	"ba":  true,
+	"jw":  true,
+	"su":  true,
+	"yue": true,
+}
+
+func normalizeTranscriptionLanguage(language string) string {
+	language = strings.ToLower(language)
+
+	if language == "auto" || language == "" {
+		return language
+	}
+
+	if ok, _ := whisperSupportedLanguages[language]; ok {
+		return language
+	}
+
+	// Try to guess the language
+	return "auto"
+}
+
 func DoTranscribe(
 	ctx context.Context,
 	inputFile string,