diff --git a/CHANGELOG.md b/CHANGELOG.md index fe2ad8e..4ba724f 100644 --- a/CHANGELOG.md +++ b/CHANGELOG.md @@ -6,6 +6,14 @@ Das Format folgt [Keep a Changelog](https://keepachangelog.com/de/1.1.0/), die V ## [Unreleased] +## [0.22.2] - 2026-10-31 + +### Fixed +- Spracheingabe über den Browser (ohne Whisper-Server): In Chrome auf Android wurde ein Satz mehrfach übernommen („TestTestTestTest 1 2 3“ statt „Test 1 2 3“), weil Android jeden Zwischenstand als eigenes Ergebnis liefert. Fortsetzungen ersetzen jetzt den vorherigen Stand. + +### Added +- `AGENTS.md` mit Regeln für Coding-Agenten (Sprache, Prüfen, Release-Ablauf über die CI) und ein Plan für Registrierung mit Freigabe/E-Mail-Endungen und Anmeldung per OpenID Connect (`Doku/plans/`). + ## [0.22.1] - 2026-10-31 ### Fixed @@ -465,7 +473,8 @@ Das Format folgt [Keep a Changelog](https://keepachangelog.com/de/1.1.0/), die V - Geteilte Listen (Einkaufs-, Pack- und Checklisten) mit `#Kategorie`-Gruppierung, Mengenangaben und Echtzeit-Abhaken. - Benutzer-, Gruppen- und Admin-Verwaltung (Ersteinrichtungs-Assistent, Einladungscodes, Profilbilder, persönliche API-Keys & Webhook-Ingest `/api/v1/webhooks/ingest`, JSON-Datenexport) sowie CSS-Theme-Engine mit 4 Themes (`Dark`, `Light`, `High Contrast`, `Soft`) und PWA-Unterstützung. -[Unreleased]: https://git.dsit.me/sandrodach/MindSift/compare/v0.22.1...dev +[Unreleased]: https://git.dsit.me/sandrodach/MindSift/compare/v0.22.2...dev +[0.22.2]: https://git.dsit.me/sandrodach/MindSift/compare/v0.22.1...v0.22.2 [0.22.1]: https://git.dsit.me/sandrodach/MindSift/compare/v0.22.0...v0.22.1 [0.22.0]: https://git.dsit.me/sandrodach/MindSift/compare/v0.21.0...v0.22.0 [0.21.0]: https://git.dsit.me/sandrodach/MindSift/compare/v0.20.0...v0.21.0 diff --git a/README.md b/README.md index 56fa12a..26cdd2e 100644 --- a/README.md +++ b/README.md @@ -6,7 +6,7 @@ [![Self-Hosted](https://img.shields.io/badge/100%25-Self--Hosted-2563eb?style=flat-square)](#) [![Docker](https://img.shields.io/badge/Deployment-Single%20Container-2496ed?style=flat-square&logo=docker&logoColor=white)](#-schnellstart) [![SQLite](https://img.shields.io/badge/Daten-SQLite-003b57?style=flat-square&logo=sqlite&logoColor=white)](#-deine-daten-gehören-dir) -[![Version](https://img.shields.io/badge/Version-v0.22.1-16a34a?style=flat-square)](CHANGELOG.md) +[![Version](https://img.shields.io/badge/Version-v0.22.2-16a34a?style=flat-square)](CHANGELOG.md) **MindSift** ist eine selbst gehostete Schaltstelle für spontane Gedanken, Aufgaben mit Kanban-Board, hierarchische Outliner-Notizen, die mehrere Personen gleichzeitig bearbeiten können, und gemeinsame Listen. Alles läuft als ein einzelner Container mit eingebetteter SQLite-Datenbank, auf deinem eigenen Server. diff --git a/frontend/package-lock.json b/frontend/package-lock.json index bca35c6..3970401 100644 --- a/frontend/package-lock.json +++ b/frontend/package-lock.json @@ -1,12 +1,12 @@ { "name": "mindsift-frontend", - "version": "0.22.1", + "version": "0.22.2", "lockfileVersion": 3, "requires": true, "packages": { "": { "name": "mindsift-frontend", - "version": "0.22.1", + "version": "0.22.2", "license": "Unlicense", "dependencies": { "@fontsource-variable/inter": "5.3.0", diff --git a/frontend/package.json b/frontend/package.json index 802eaad..a6389c1 100644 --- a/frontend/package.json +++ b/frontend/package.json @@ -1,7 +1,7 @@ { "name": "mindsift-frontend", "private": true, - "version": "0.22.1", + "version": "0.22.2", "license": "Unlicense", "type": "module", "scripts": { diff --git a/frontend/src/lib/__tests__/voice.join.test.ts b/frontend/src/lib/__tests__/voice.join.test.ts new file mode 100644 index 0000000..c4118ce --- /dev/null +++ b/frontend/src/lib/__tests__/voice.join.test.ts @@ -0,0 +1,21 @@ +import { describe, it, expect } from 'vitest'; +import { joinSpeechResults } from '../voice'; + +describe('joinSpeechResults (browser speech recognition)', () => { + it('Android: growing interim states of one phrase count once', () => { + expect(joinSpeechResults(['Test', 'Test', 'Test', 'Test', 'Test', 'Test 1 2 3'])).toBe('Test 1 2 3'); + expect(joinSpeechResults(['Test', 'Test 1', 'Test 1 2', 'Test 1 2 3'])).toBe('Test 1 2 3'); + }); + + it('desktop: consecutive pieces are joined', () => { + expect(joinSpeechResults(['Milch kaufen', ' und Brot'])).toBe('Milch kaufen und Brot'); + }); + + it('Android: a new phrase after a finished one is appended', () => { + expect(joinSpeechResults(['Milch', 'Milch kaufen', ' und Brot', ' und Brot holen'])).toBe('Milch kaufen und Brot holen'); + }); + + it('ignores empty results and earlier shorter states', () => { + expect(joinSpeechResults(['', 'Test 1 2', 'Test 1', ''])).toBe('Test 1 2'); + }); +}); diff --git a/frontend/src/lib/useVoiceInput.ts b/frontend/src/lib/useVoiceInput.ts index 783d683..f44fe75 100644 --- a/frontend/src/lib/useVoiceInput.ts +++ b/frontend/src/lib/useVoiceInput.ts @@ -1,6 +1,6 @@ import { useCallback, useEffect, useRef, useState } from 'react'; import { api } from './api'; -import { browserRecognition, currentVoiceConfig, EffectiveVoiceMode, loadVoiceConfig, VoiceRecorder } from './voice'; +import { browserRecognition, currentVoiceConfig, EffectiveVoiceMode, joinSpeechResults, loadVoiceConfig, VoiceRecorder } from './voice'; export type VoiceState = 'idle' | 'listening' | 'transcribing'; @@ -234,8 +234,9 @@ export function useVoiceInput({ onText, onError, continuous = false, live = fals recognition.interimResults = true; recognition.continuous = continuous || live; recognition.onresult = (event: any) => { - let sessionText = ''; - for (let i = 0; i < event.results.length; i++) sessionText += event.results[i][0].transcript; + const parts: string[] = []; + for (let i = 0; i < event.results.length; i++) parts.push(event.results[i][0].transcript); + const sessionText = joinSpeechResults(parts); if (live && baseTranscriptRef.current) { const base = baseTranscriptRef.current.trimEnd(); const next = sessionText.trimStart(); diff --git a/frontend/src/lib/voice.ts b/frontend/src/lib/voice.ts index 8e447f4..40a4903 100644 --- a/frontend/src/lib/voice.ts +++ b/frontend/src/lib/voice.ts @@ -186,3 +186,27 @@ export class VoiceRecorder { this.analyser = null; } } + +/** + * Text of one speech recognition session from its results. Desktop browsers deliver consecutive + * pieces ("Test 1 2 3", " und weiter"); Chrome on Android also keeps the growing interim states of + * the same phrase as results of their own ("Test", "Test", "Test 1 2 3"). A result that only + * continues the one before it therefore replaces it instead of being appended. + */ +export function joinSpeechResults(transcripts: string[]): string { + const norm = (s: string) => s.trim().toLowerCase(); + let text = ''; + let last = ''; + for (const t of transcripts) { + if (!t) continue; + if (last && norm(t).startsWith(norm(last))) { + text = text.slice(0, text.length - last.length) + t; // grown version of the same phrase + } else if (last && norm(last).startsWith(norm(t))) { + continue; // an earlier, shorter state of it + } else { + text += t; + } + last = t; + } + return text; +}