From 90dfe73f1a1b6815cfba08ec165394719005e4cf Mon Sep 17 00:00:00 2001
From: Ariver <shanghai3168@gmail.com>
Date: Sun, 07 Jun 2026 01:48:32 +0800
Subject: [PATCH] Fix X-ASR CJK spacing
---
privatevoice.src/app.go | 2 +-
privatevoice.src/internal/textproc/textproc_test.go | 27 +++++++++++++++++++++++++++
CHANGELOG.md | 2 +-
privatevoice.src/frontend/src/components/settings/AboutPage.svelte | 2 +-
privatevoice.src/internal/textproc/textproc.go | 25 +++++++++++++++++++++++++
privatevoice.src/build/darwin/Info.plist | 2 +-
6 files changed, 56 insertions(+), 4 deletions(-)
diff --git a/CHANGELOG.md b/CHANGELOG.md
index 736069d..1a3f527 100755
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -11,7 +11,7 @@
### 构建
-- build: `20260607.0141`
+- build: `20260607.0147`
- 说明: 本版本为 X-ASR 技术实验候选包，不是 App Store 正式上传包。
---
diff --git a/privatevoice.src/app.go b/privatevoice.src/app.go
index 9e06899..d4e1ccd 100755
--- a/privatevoice.src/app.go
+++ b/privatevoice.src/app.go
@@ -28,7 +28,7 @@
const (
appVersion = "2.1.29"
- appBuild = "20260607.0141"
+ appBuild = "20260607.0147"
appDisplayVersion = appVersion + " (build " + appBuild + ")"
appName = "PrivateVoice Dictation"
diff --git a/privatevoice.src/build/darwin/Info.plist b/privatevoice.src/build/darwin/Info.plist
index 14cae8f..4ea4a40 100755
--- a/privatevoice.src/build/darwin/Info.plist
+++ b/privatevoice.src/build/darwin/Info.plist
@@ -19,7 +19,7 @@
<key>CFBundleShortVersionString</key>
<string>2.1.29</string>
<key>CFBundleVersion</key>
- <string>20260607.0141</string>
+ <string>20260607.0147</string>
<key>LSApplicationCategoryType</key>
<string>public.app-category.productivity</string>
<key>ITSAppUsesNonExemptEncryption</key>
diff --git a/privatevoice.src/frontend/src/components/settings/AboutPage.svelte b/privatevoice.src/frontend/src/components/settings/AboutPage.svelte
index 252385b..f541db4 100644
--- a/privatevoice.src/frontend/src/components/settings/AboutPage.svelte
+++ b/privatevoice.src/frontend/src/components/settings/AboutPage.svelte
@@ -3,7 +3,7 @@
import { t } from '../../lib/i18n'
import AppIcon from '../shared/AppIcon.svelte'
- let version = $state('2.1.29 (build 20260607.0141)')
+ let version = $state('2.1.29 (build 20260607.0147)')
const currentYear = new Date().getFullYear().toString()
const aboutLinks = [
{ labelKey: 'about.sourceProjectLabel', urlKey: 'about.sourceUrl' },
diff --git a/privatevoice.src/internal/textproc/textproc.go b/privatevoice.src/internal/textproc/textproc.go
index 44d9cdb..60683de 100755
--- a/privatevoice.src/internal/textproc/textproc.go
+++ b/privatevoice.src/internal/textproc/textproc.go
@@ -59,8 +59,31 @@
return strings.TrimSpace(text)
}
+func removeSpacesBetweenCJK(text string) string {
+ runes := []rune(text)
+ buf := make([]rune, 0, len(runes))
+
+ for i := 0; i < len(runes); i++ {
+ r := runes[i]
+ if unicode.IsSpace(r) {
+ j := i
+ for j < len(runes) && unicode.IsSpace(runes[j]) {
+ j++
+ }
+ if len(buf) > 0 && j < len(runes) && isCJK(buf[len(buf)-1]) && isCJK(runes[j]) {
+ i = j - 1
+ continue
+ }
+ }
+ buf = append(buf, r)
+ }
+
+ return string(buf)
+}
+
// PostProcess cleans up mixed Chinese-English text from SenseVoice:
// - Strips special tokens (<|zh|>, <|NEUTRAL|>, etc.)
+// - Removes model-added spaces between CJK characters
// - Inserts spaces between CJK characters and ASCII letters/digits
// - Capitalizes the first letter and letters after sentence-ending punctuation
// - Collapses multiple spaces
@@ -78,6 +101,8 @@
return text
}
+ text = removeSpacesBetweenCJK(text)
+
// Insert spaces at CJK ↔ ASCII alphanumeric boundaries
runes := []rune(text)
buf := make([]rune, 0, len(runes)+16)
diff --git a/privatevoice.src/internal/textproc/textproc_test.go b/privatevoice.src/internal/textproc/textproc_test.go
new file mode 100644
index 0000000..aaa0be7
--- /dev/null
+++ b/privatevoice.src/internal/textproc/textproc_test.go
@@ -0,0 +1,27 @@
+package textproc
+
+import "testing"
+
+func TestPostProcessRemovesSpacesBetweenCJK(t *testing.T) {
+ got := PostProcess("请 你 根 据 官 网")
+ want := "请你根据官网"
+ if got != want {
+ t.Fatalf("PostProcess() = %q, want %q", got, want)
+ }
+}
+
+func TestPostProcessKeepsSpaceBetweenCJKAndEnglish(t *testing.T) {
+ got := PostProcess("请 你 open AI")
+ want := "请你 open AI"
+ if got != want {
+ t.Fatalf("PostProcess() = %q, want %q", got, want)
+ }
+}
+
+func TestPostProcessKeepsEnglishWordSpaces(t *testing.T) {
+ got := PostProcess("open source speech recognition")
+ want := "Open source speech recognition"
+ if got != want {
+ t.Fatalf("PostProcess() = %q, want %q", got, want)
+ }
+}
--
Gitblit v1.9.3