輸入 1–5 首參考歌曲，分析曲風後生成近似風格的原創配樂。全部在瀏覽器內執行，音樂檔不會上傳到任何伺服器。Load 1–5 reference songs, analyse their style and generate original music in a similar style. Everything runs in your browser; your audio is never uploaded.
⚙設定Settings
自訂按鍵Custom keys
點選按鍵欄位後按下新按鍵即可設定；Esc 取消、Backspace 清除。設定保存在此瀏覽器。Click a key box, then press the new key. Esc cancels, Backspace clears. Settings are saved in this browser.
暫存空間管理Storage
本工具佔用：This tool uses:
音訊快取（原曲、伴奏、人聲）Audio cache (songs, instrumental, vocals): 音色庫快取Instrument bank cache: 分析資料與工作階段Analysis data and session:
1輸入歌曲Input songs
音訊快取：Audio cache:
支援 MP3 / WAV / M4A / OGG / FLAC，每首最長處理 10 分鐘。權重越高，該曲對風格的影響越大。讀入分析檔可直接略過第 2 步。MP3 / WAV / M4A / OGG / FLAC, up to 10 minutes per song. A higher weight gives that song more influence on the style. Loading an analysis file skips step 2.
2去除人聲（批次佇列）Vocal removal (batch queue)
儲存Save
逐首處理以節省記憶體：先以立體聲比對快速分離，伴奏用於分析和聲、律動與樂器，人聲用於擷取旋律特徵。AI 分離首次需下載約 67 MB 模型（之後快取），速度視電腦效能而定。選「不去人聲」時直接以原曲分析（適合純音樂或伴奏版），速度最快。可在每首歌的波形上拖曳，只分析其中一段（例如戰鬥高潮），最長 10 分鐘。分析完成後伴奏與人聲以原始取樣率保留，可試聽，或按每首歌的「儲存」選擇格式後下載；來自分析檔的歌曲按試聽或儲存時，會請你選擇原曲並自動去人聲；勾選「在瀏覽器暫存」時，原曲與伴奏 / 人聲會存在本機瀏覽器，重新整理後自動還原。Songs are processed one at a time to save memory: a fast stereo split first; the instrumental feeds harmony, groove and instrument analysis, the vocals feed melody features. AI separation downloads a ~67 MB model once (cached); speed depends on the computer. Choose No vocal removal to analyse the original mix directly (best for instrumentals; fastest). Drag on each song's waveform to analyse only part of it (e.g. the battle climax), up to 10 minutes. The instrumental and vocals are kept at the original sample rate for listening, or press Save on a song to pick the format and download; for songs from an analysis file you are asked for the original and the vocals are removed automatically; with the browser cache on, the songs and their parts are stored locally and come back after a reload.
3曲風分析與比較Style analysis & comparison
信心度低（紅 / 黃）時請確認：可能是半速 / 倍速、關係大小調混淆或拍號誤判。可直接修改 BPM、調性與拍號，修正後會重新計算並存入分析檔。Low confidence (red / yellow) means please check: half / double tempo, relative major / minor or a wrong meter are common. You can edit BPM, key and meter; corrections are recomputed and stored in the analysis file.
兩兩相似度Pairwise similarity
合併後的風格檔案Merged style profile
曲風標籤與樂器預設以頻譜特徵啟發式判斷；選用 AI 辨識（YAMNet）時會結合模型結果（標示「AI」）。Genre tags and instruments are heuristic estimates from spectral features; with AI tagging (YAMNet) the model results are blended in (marked "AI").
分析檔為 JSON 純文字（.aemlstyle.json），不含任何音訊；預設只存統計與原創性指紋，不含原曲旋律。開啟「保存完整旋律音符」後原曲旋律會寫入檔案，分享前請留意原曲著作權。The analysis file is plain JSON (.aemlstyle.json) without any audio; by default it holds statistics and originality fingerprints, not the original melodies. With "save the full melody notes" on, the original melodies are written into the file; mind the copyright of the originals before sharing it.
4作曲參數Composition parameters
（選一個場景即一次設定 8 項參數；設為「自動」的項目才會被場景覆寫）(a scene sets all 8 parameters at once; only items left on Auto are overridden)
長度Length
–
節奏Rhythm
旋律Melody
和聲Harmony
主律動Foundation
旋律線Lead
襯底音Pad
填音Fill
編制層次Arrangement
旋律、和聲與音色細節Melody, harmony and sound details
AI 旋律模型為 Google Magenta 的 ImprovRNN（依和弦即興的 LSTM 神經網路，約 5 MB，加上程式庫約 2.5 MB，首次使用時下載）。它依本工具產生的和弦進行寫出主旋律，其餘聲部仍由編曲引擎完成。目前沒有可在瀏覽器執行、50 MB 以下的公開 Music Transformer 模型，因此採用這個體積小、以和弦為條件的模型。The AI melody model is Google Magenta's ImprovRNN (a chord-conditioned LSTM, ~5 MB plus a ~2.5 MB library, downloaded on first use). It writes the lead over the chords this tool generates; the arrangement engine writes the other parts. There is no public browser-ready Music Transformer under 50 MB, so this small chord-conditioned model is used instead.
自訂格式：以 - 分隔段落，例如 Intro-A-A-B-A-Outro；可用 A:4 指定該段小節數。自訂結構時長度由結構決定。Custom format: sections separated by -, e.g. Intro-A-A-B-A-Outro; A:4 sets that section's bar count. With a custom structure the length follows the structure.
主旋律 MIDIMelody MIDI
匯入後主旋律固定使用你的 MIDI（取最高聲部、量化到 16 分音符），工具依旋律自動配和弦、貝斯、鼓與其他聲部；速度、拍號、調性取自 MIDI，除非另行指定。The imported melody (top voice, quantised to 16ths) is kept; the tool harmonises it and writes the bass, drums and other parts. Tempo, meter and key come from the MIDI unless you set them.
5生成與試聽Generate & preview
0:00 / 0:00
樂器音色（真實取樣）Instrument sounds (real samples)
以真實樂器錄音取樣取代合成音色：General MIDI 音色庫（MusyngKite / FluidR3）、Salamander 平台鋼琴、取樣鼓組。第一次使用需下載（每種樂器約 0.3–1.5 MB），之後存在瀏覽器、可離線使用。試聽與匯出使用同一組音色。Recorded instrument samples replace the synth voices: General MIDI libraries (MusyngKite / FluidR3), the Salamander grand piano and sampled drum kits. Downloaded once (about 0.3–1.5 MB per instrument), then kept in the browser and usable offline. Preview and export use the same sounds.
本曲各樂器的音色（可換成同類的其他樂器；綠色＝真實取樣、灰色＝合成器）：Sound of each instrument in this song (swap for a related instrument; green = sampled, grey = synth):
鼓組取樣含大鼓、小鼓、Hi-hat、筒鼓；銅鈸與手鼓類仍為合成。音色檔案來源：gleitz/midi-js-soundfonts、Tone.js audio（Salamander、drum-samples）。Drum kits sample kick, snare, hi-hat and toms; cymbals and hand percussion stay synthesized. Sources: gleitz/midi-js-soundfonts, Tone.js audio (Salamander, drum-samples).
多軌聲部圖與音符編輯Multi-track view and note editor
上方：各聲部即時發聲狀態（亮燈＝正在發聲，點一列可選取該聲部並跳到該處）。下方：所選聲部的鋼琴卷簾。Top: which parts are sounding right now (lit = sounding; click a row to pick that part and jump there). Bottom: piano roll of the chosen part.
點音符選取，上下拖曳改音高、左右拖曳移位置（對齊 16 分音符）；雙擊空白處新增音符；Delete 刪除、方向鍵微調、Ctrl+Z 復原；紅框＝調外音（可能走音）。修改會立即重新算繪並套用到匯出與 MIDI。Click a note to select; drag up / down for pitch, left / right to move (16th grid); double-click empty space to add; Delete removes, arrow keys nudge, Ctrl+Z undoes; red outline = outside the key (may sound off). Edits are re-rendered at once and used for export and MIDI.
段落編排表Section arrangement grid
逐段開關各聲部（取消勾選＝該段不演奏），變更後自動重新算繪。Switch each part on or off per section (unticked = silent there); re-renders automatically.
鎖定的聲部在「重新生成」時保持不變；鎖定旋律、和聲、貝斯或襯底音時和弦進行也會一併保留。Locked parts stay unchanged on "Regenerate"; locking the lead, harmony, bass or pad also keeps the chord progression.
局部重生Partial regeneration
生成記錄History
保留最近 12 次結果：可載入、設為 B 做 A/B 比較，或加入最愛（最愛不會被自動清除）。The last 12 results are kept: load one, set it as B for A/B comparison, or mark it as a favourite (favourites are never auto-removed).
依目前混音台的靜音 / 獨奏與段落編排表輸出；鼓組在第 10 聲道（GM 鼓組對照），含速度、拍號、調號、段落標記與循環點。Follows the mixer mute / solo and the section grid; drums on channel 10 (GM drum map), with tempo, meter, key, section markers and loop points.
全部在瀏覽器內編碼。開啟無縫循環時，循環點寫入 WAV 的 smpl 區塊或 OGG / FLAC 的 LOOPSTART 標籤，並另存一份循環點文字檔（_loop.txt）。檔案中繼資料會寫入作者與工具名稱。輸入歌曲的著作權屬原作者；本工具只學習風格，並以原創性檢查避免旋律雷同。Encoded entirely in your browser. With a seamless loop the loop points go into the WAV smpl chunk or the OGG / FLAC LOOPSTART tags, plus a loop-point text file (_loop.txt). Author and tool name are written into the file metadata. The input songs remain the property of their authors; the tool learns only the style and an originality check prevents melodic copies.
7輸出樂譜Sheet music
0:00 / 0:00
生成的曲子本身就是音符資料，樂譜與 MIDI 完全精確。MusicXML 為多聲部（含鼓譜），可用 MuseScore、Finale、Sibelius 開啟；多音軌 MIDI 的鼓組在第 10 聲道，可直接匯入 DAW 或遊戲引擎。The generated piece is note data, so score and MIDI are exact. MusicXML is multi-part (including a drum staff) and opens in MuseScore, Finale or Sibelius; the multi-track MIDI keeps drums on channel 10 for DAWs and game engines.
點擊選擇或拖曳音樂檔Click or drop an audio file歌曲Song權重Weight等待分析Waiting已分析Analysed來自分析檔From analysis file失敗Failed片段Segment（音訊來自瀏覽器暫存）(audio from the browser cache)已清除暫存音訊。Cached audio cleared.已從瀏覽器暫存還原音訊，可直接試聽與重新分析。Audio restored from the browser cache; you can listen and re-analyse right away.載入 AI 辨識模型（YAMNet）...Loading the AI tagging model (YAMNet)...AI 辨識樂器與曲風中...AI tagging instruments and style...AI 辨識完成（YAMNet）：AI tagging done (YAMNet):無法使用 AI 辨識模型（可能離線），已改用啟發式：The AI tagging model is unavailable (offline?); using the heuristic instead:偵測調性中...Detecting the key...嘻哈hip-hop電子 / EDMEDMR&B / 靈魂R&B / soul爵士 / 藍調jazz / blues古典classical民謠 / 世界folk / world遊戲配樂game music分析片段Segment to analyse在波形上拖曳選取（最長 10 分鐘），點一下可跳到該處試聽Drag on the waveform to select (max 10 min); click to listen from there整首Whole song片段已變更，請重新分析。Segment changed; analyse again.這首歌沒有音訊（來自分析檔），請選擇它的原曲音檔...This song has no audio (from an analysis file); choose its original audio file...已連結原曲：Original linked:已連結，但檔案與分析時的原曲不同（檢查碼不符），結果可能不一致：Linked, but this file differs from the analysed original (hash mismatch); results may differ:去人聲中（產生伴奏與人聲）...Removing vocals (making the instrumental and vocals)...伴奏與人聲已產生Instrumental and vocals ready去人聲失敗：Vocal removal failed:分析進行中，請稍候再試。Analysis is running; try again when it finishes.重新開始Start over確定清除全部歌曲與結果？再按一次Clear all songs and results? Press again已重新開始：歌曲、分析、生成結果與其暫存音訊已清除（參數與設定保留）。可以放入新的歌曲了。Started over: songs, analysis, results and their cached audio are cleared (parameters and settings kept). Add new songs.已清除音色庫快取（下次使用時重新下載）。Instrument cache cleared (downloaded again when needed).納入Use沒有勾選任何歌曲：生成將使用預設風格。請在第一欄勾選要納入的歌曲。No song is ticked: generation uses the default style. Tick the songs to use in the first column.點音符選取，上下拖曳改音高、左右拖曳移位置（對齊 16 分音符）；雙擊空白處新增音符；Delete 刪除；紅色＝調外音（可能走音）。Click a note to select; drag up / down for pitch, left / right to move (16th grid); double-click empty space to add; Delete removes; red = outside the key (may sound off).已修正調外音：Out-of-key notes fixed:沒有調外音。No out-of-key notes.已套用修改並重新算繪。Edit applied and re-rendered.音量Volume平移Pan殘響Reverb自動（編曲預設值）Auto (the arrangement's setting)AI 模型依和弦寫旋律中...The AI model is writing a melody over the chords...AI 旋律完成（候選曲中標示 AI）。AI melody done (marked AI in the candidates).無法使用 AI 旋律模型（可能離線），已改用規則引擎The AI melody model is unavailable (offline?); using the rule engine已載入 IR：IR loaded:無法讀取這個 IR 檔。Could not read this IR file.儲存Save請先勾選要儲存的伴奏或人聲（需先完成去人聲）。Tick the instrumental and / or vocals to save (vocal removal must be done first).編碼中：Encoding:已儲存：Saved:逐段開關各聲部（取消勾選＝該段不演奏），變更後自動重新算繪。Switch each part on or off per section (unticked = silent there); re-renders automatically.全部開啟All on已匯入主旋律：Melody imported:無法讀取這個 MIDI 檔（找不到旋律音符）。Could not read this MIDI file (no melody notes found).（未匯入，主旋律自動生成）(none; the melody is generated)已選取小節：Selected bars:已更換樂器，重新算繪中...Instrument changed; re-rendering...解碼音訊中...Decoding audio...去除人聲中...Removing vocals...AI 人聲分離中...AI vocal separation...分析特徵中...Analysing features...辨識和弦中...Recognising chords...擷取旋律中...Extracting melody...分析完成！已建立風格檔案。Analysis done! Style profile created.已取消。Cancelled.無法解碼此檔案，請換成 MP3 或 WAV。Could not decode this file. Try MP3 or WAV.超過 10 分鐘，僅處理前 10 分鐘。Longer than 10 minutes; only the first 10 minutes are processed.請先放入至少一首歌曲。Add at least one song first.最多 5 首。Up to 5 songs.歌曲Song狀態Stage進度Progress分離Split快速fastAIAI不去人聲none試聽Listen播放Play停止Stop原曲Original伴奏Instrumental人聲Vocals（無音訊）(no audio)此檔案無法在瀏覽器中播放。This file cannot be played in the browser.AI 模型載入中...Loading the AI model...AI 模型下載中（首次約 67 MB，之後會快取）：Downloading the AI model (~67 MB, first time only):AI 人聲分離中：AI separation:AI 分離完成AI separation done無法使用 AI 模型（可能離線或瀏覽器不支援），已改用快速分離：The AI model is unavailable (offline or unsupported browser); using the fast split instead:歌曲Song長度Length速度 BPMTempo BPM調性Key拍號Meter亮度Brightness動態範圍Dynamics響度Loudness主要樂器Main instruments曲風標籤Genre tags段落Form旋律Melody權重Weight大調major小調minor大調major自然小調natural minor和聲小調harmonic minor多利安Dorian利底亞Lydian弗里吉亞Phrygian五聲音階pentatonic半音chromatic可能是半速maybe half tempo可能是倍速maybe double tempo可能是關係大小調maybe the relative key已手動修正corrected以下歌曲風格偏離其他歌曲：These songs deviate from the others in style:（已自動降低權重；可在「合併策略」改為剔除或保留）(automatically down-weighted; change the merge strategy to drop or keep them)（已剔除）(dropped)（已降低權重）(down-weighted)（仍納入合併）(still merged)速度Tempo主調Key拍號Meter平均長度Average length能量Energy搖擺Swing換和弦 / 小節Chord changes / bar旋律密度（音 / 拍）Melody density (notes / beat)旋律音域Melody range鼓組比重Drums段落模板Form template曲風標籤Genre tags輸入歌曲的和弦轉移（Markov）input chord transitions (Markov)合併風格Merged原創性指紋Originality fingerprints常見和弦Common chords速度Tempo能量Energy亮度Brightness動態Dynamics鼓組Drums和聲變化Harmony音域Range搖擺Swing管弦 / 氛圍orchestral / ambient搖滾rock電子electronic抒情ballad搖擺swing快節奏fast氛圍ambient流行pop已下載分析檔：Analysis file saved:已讀入分析檔，歌曲數：Analysis file loaded. Songs:檔案不是有效的 JSON。The file is not valid JSON.這不是本工具的分析檔。This is not an analysis file of this tool.分析檔版本較新，請更新本工具。The analysis file is from a newer version; please update the tool.分析檔缺少必要欄位，可能已損毀。Required fields are missing; the file may be damaged.檢查碼不符：檔案已損毀或被修改，為確保品質已停止讀入。Checksum mismatch: the file is damaged or was modified; loading stopped to protect output quality.數值超出合理範圍：Value out of range:已將舊版分析檔自動轉換。An older analysis file was converted automatically.注意：作曲引擎版本與建檔時不同，輸出可能與原先不同。Note: the composition engine version differs from the one that made the file; output may differ.可再放入新歌曲並按「開始分析」，與分析檔內的歌曲一起合併。You can add new songs and press Analyse to merge them with the songs from the file.這不是本工具的作品檔。This is not a project file of this tool.已讀入作品檔：Project file loaded:偵測到裝置記憶體或處理器核心較少：已自動減少候選曲數量並停用 AI 分離，可在「效能模式」改為高品質。This device has little memory or few CPU cores: fewer candidates and no AI separation are used automatically. Switch "Performance" to High quality to override.繪圖：WebGPU 硬體加速Rendering: WebGPU hardware繪圖：軟體（Canvas 2D）Rendering: software (Canvas 2D)此瀏覽器不支援 WebGPU，已自動改用軟體繪圖。This browser has no WebGPU; switched to software rendering.此瀏覽器沒有 WebGPU（需 Chrome / Edge 113 以上），已自動改用軟體繪圖。This browser has no WebGPU (needs Chrome / Edge 113+); switched to software rendering.網頁不是以 https 開啟，瀏覽器不開放 WebGPU，已自動改用軟體繪圖。請改用 https:// 網址。The page is not served over https, so the browser hides WebGPU; switched to software rendering. Open the https:// address instead.瀏覽器有 WebGPU，但取不到顯示卡（可能關閉了「使用圖形加速功能」，或顯示卡驅動被 Chrome 停用；可在 chrome://gpu 查看），已自動改用軟體繪圖。The browser has WebGPU but no GPU adapter (hardware acceleration may be off or the driver blocked; see chrome://gpu); switched to software rendering.畫布無法切換為 WebGPU，已自動改用軟體繪圖；請重新整理頁面再試。The canvas could not switch to WebGPU; switched to software rendering. Reload the page and try again.顯示卡的 WebGPU 連線中斷（可能是驅動程式重置），已自動改用軟體繪圖；重新整理頁面可再試 WebGPU。The WebGPU device was lost (possibly a driver reset); switched to software rendering. Reload the page to try WebGPU again.WebGPU 已啟動但畫面沒有內容，已自動改用軟體繪圖。WebGPU started but drew nothing; switched to software rendering.WebGPU 初始化失敗，已自動改用軟體繪圖。WebGPU failed to start; switched to software rendering.此瀏覽器不支援原生全螢幕，已改用頁面內全螢幕。Native full screen is not supported here; using in-page full screen.請按下按鍵...Press a key...（未設定）(none)已還原上次的工作階段。Your previous session was restored.偵測到程式碼遭修改，本工具已停用。Source modification detected. This tool has been disabled.尚無風格檔案，將使用預設的流行風格。建議先分析歌曲或讀入分析檔。No style profile yet; a default pop style is used. Analyse songs or load an analysis file first for best results.生成候選曲中...Generating candidates...合成音訊中...Rendering audio...完成！最高分候選曲已載入，可試聽、切換其他候選曲或匯出。Done! The best candidate is loaded; listen, switch candidates or export.候選Candidate總分Score風格相似style和聲協和consonance旋律平順smoothness原創性originality種子seed原創性檢查通過：未發現與輸入歌曲連續 8 音相同的旋律。Originality check passed: no 8-note run identical to the input songs.原創性檢查：曾發現雷同片段並已自動重生，次數：Originality check: similar passages were found and regenerated automatically, times:無輸入歌曲的旋律指紋，未進行原創性比對。No melody fingerprints from input songs; originality comparison skipped.警告：仍有與輸入歌曲相同的旋律片段，建議重新生成。片段數：Warning: some melodic passages still match the inputs; please regenerate. Passages:旋律Lead和聲Harmony貝斯Bass鼓組Drums襯底音Pad填音Fill副旋律Counter-melody琶音Arpeggio旋律和聲Lead harmony打擊樂Percussion靜音M獨奏S鎖定Lock載入Load設為 BSet B目前播放：A（目前作品）Playing: A (current)目前播放：BPlaying: B請先在生成記錄中按「設為 B」。Set a history item as B first.局部重生完成。Partial regeneration done.小節範圍不正確。Invalid bar range.循環接縫試聽中（結尾 4 秒 → 開頭），再按一次停止。Auditioning the loop seam (last 4 s → start); press again to stop.請先在「長度」開啟無縫循環並重新生成。Turn on the seamless loop under Length and regenerate first.自動 →auto →循環loop秒s小節bars編碼中...Encoding...匯出完成：Exported:個檔案files請先生成作品。Generate a piece first.此瀏覽器無法編碼所選格式，已改存為 WAV。This browser cannot encode the chosen format; saved as WAV instead.強度分層intensity layer批次生成中：Batch generating:下載真實樂器取樣中Downloading instrument samples使用真實取樣的樂器：Instruments using real samples:載入 SoundFont 音色中...Loading SoundFont instruments...SoundFont 無法載入（可能離線），已改用合成器音色。SoundFont could not load (offline?); using the synth instead.多個檔案已打包成 ZIP：Several files were packed into a ZIP:短樂句stinger已儲存作品檔：Project file saved:載入 MP3 編碼元件...Loading the MP3 encoder...編碼中...Encoding...此瀏覽器需即時錄製編碼，所需時間等於音樂長度，請稍候...This browser encodes in real time (takes the track length), please wait...請先生成作品。Generate a piece first.樂譜繪製元件無法載入（網路受限），已改以簡易譜面顯示；下載不受影響。The engraver could not load (network); a simple view is shown instead. Downloads still work.樂譜完成。Score ready.試聽範例中...Playing an example...本版本為階段性交付。This is a staged delivery.
採用輸入歌曲的平均長度。Uses the average length of the input songs.在指定區間內依種子挑選長度，並對齊小節。Picks a length inside the range (by seed), aligned to bars.依分析出的拍號與搖擺感決定。Follows the analysed meter and swing.最常見的四拍，穩定直進。The common straight four-beat feel.三拍子，圓舞曲感，適合城鎮、童話。Three beats, waltz-like; towns, fairy tales.兩大拍、每拍三分，搖晃流動，適合航海、民謠。Two big beats split in three; rolling, for sea or folk scenes.反拍延後成三連音位置，爵士 / 輕鬆感。Off-beats delayed toward triplets; jazzy and relaxed.重音落在反拍，活潑有推進力。Accents on off-beats; lively and driving.小鼓只在第 3 拍，感覺放慢一半、厚重。Snare on beat 3 only; feels half as fast and heavy.小鼓加倍，感覺速度加倍、緊迫。Doubled snare; feels twice as fast and urgent.整齊的進行曲小鼓，適合勝利、軍隊。Tight march snare; victory and armies.無鼓、無明顯拍點，氛圍流動。No drums or clear pulse; floating ambience.採用分析出的調式。Uses the analysed mode.明亮、正面。Bright and positive.憂傷、內斂。Sad and introspective.五聲音階，東方、民謠色彩，不易出錯。Five notes; eastern / folk colour and very safe.升高第七音，神秘、異域。Raised 7th; mysterious and exotic.小調但第六音明亮，冒險、探索。Minor with a bright 6th; adventure and exploration.大調加升四級，夢幻、奇幻。Major with a raised 4th; dreamy fantasy.降二級，緊張、沙漠感。Flat 2nd; tense, desert-like.加入半音經過音，不安、恐怖。Chromatic passing notes; uneasy, horror.依輸入歌曲的和弦轉移機率（Markov）產生。Generated from the chord-transition probabilities (Markov) of the inputs.I–V–vi–IV，最常見的流行進行。I–V–vi–IV, the classic pop loop.i–VI–III–VII，壯闊、史詩。i–VI–III–VII, wide and epic.ii–V–I 七和弦，成熟、都會。ii–V–I with sevenths; mellow and urban.I–IV–V–I 與終止式，端正。I–IV–V–I with cadences; formal.兩個和弦來回，調式色彩，適合選單、循環。Two chords back and forth; modal colour for menus and loops.低音持續主音，上方和弦變化，張力漸增。Bass holds the tonic while chords move above; rising tension.減和弦、增和弦與半音下行低音，懸疑。Diminished / augmented chords and a chromatic bass; suspense.依輸入的鼓組與貝斯特徵決定。Follows the drum and bass features of the inputs.大鼓 1、3 拍，小鼓 2、4 拍，八分音符貝斯。Kick on 1 and 3, snare on 2 and 4, eighth-note bass.每拍大鼓、反拍開放鈸，電子舞曲。Kick on every beat, open hats on off-beats; dance music.噪音鼓與八度跳動貝斯，復古遊戲機。Noise drums and octave-hopping bass; retro consoles.慵懶切分的大鼓、柔和小鼓。Lazy syncopated kick and soft snare.定音鼓式低音鼓與長音貝斯（大提琴）。Timpani-like hits and long cello bass.克拉維節奏、康加鼓與 tumbao 貝斯。Clave, congas and tumbao bass.密集大鼓與 16 分音符，緊張戰鬥感。Busy kick and 16th notes; tense battle drive.只有貝斯，沒有鼓組。Bass only, no drums.流行鼓組（拍手＋小鼓在 2、4 拍），切分的貝斯，鋼琴八分音符和弦。Pop kit (clap + snare on 2 and 4), syncopated bass, eighth-note piano chords.雙大鼓八分音符、重擊小鼓，破音吉他強力和弦悶音。Double-kick eighths, heavy snare, palm-muted power chords on distorted guitar.Boom-bap 鼓（略帶搖擺），808 低音跟著大鼓，電鋼琴和弦。Boom-bap drums (slight swing), 808 sub bass following the kick, electric-piano chords.慢速律動與鬼音小鼓，16 分音符腳踏鈸，電鋼琴七和弦。Laid-back groove with ghost notes, 16th hi-hats, electric-piano seventh chords.搖擺 Ride、行走貝斯、鋼琴無根音七和弦伴奏（Comping）。Swing ride, walking bass, rootless seventh-chord piano comping.輕大鼓與鼓框、沙鈴鈴鼓，木吉他刷弦，原聲貝斯。Light kick and rim, shaker and tambourine, strummed acoustic guitar, acoustic bass.無鼓、無貝斯。No drums, no bass.採用輸入中偵測到的主奏樂器。Uses the lead instrument detected in the inputs.鋼琴，通用、溫暖。Piano; versatile and warm.方波，8-bit 復古遊戲音色。Square wave; 8-bit retro.合成主奏，明亮帶顫音。Synth lead; bright with vibrato.小提琴，抒情、歌唱性。Violin; lyrical and singing.長笛，輕盈、自然。Flute; light and natural.銅管，英勇、號角。Brass; heroic fanfare.電吉他，搖滾。Electric guitar; rock.木吉他，田園、民謠。Acoustic guitar; pastoral folk.中音薩克斯風，爵士與 R&B 常用。Alto sax; common in jazz and R&B.音樂盒，童趣或詭異。Music box; childlike or eerie.依輸入的弦樂 / 合成 Pad 比重決定。Follows the strings / synth-pad balance of the inputs.弦樂長音鋪底。Sustained string bed.柔和的合成 Pad。Soft synth pad.風琴長音，莊嚴或復古。Organ chords; solemn or retro.合唱 Aah，神聖、史詩。Choir aah; sacred or epic.主音與五度持續音，環境、恐怖。Tonic + fifth drone; ambient or horror.不使用襯底音。No pad.依能量決定頻率，混合鼓過門、Crash 與琶音。Frequency by energy; mixes drum fills, crashes and arpeggios.每 4 小節一個過門。A fill every 4 bars.每 8 小節一個過門。A fill every 8 bars.只在段落轉換時過門。Fills only at section changes.中鼓 / 落地鼓過門。Tom fills.上行琶音帶入下一段。A rising arpeggio into the next section.噪音掃頻上升，製造期待。A rising noise sweep that builds anticipation.只在段落開頭敲 Crash 鈸。Only a crash cymbal at section starts.依輸入歌曲的樂器數與能量決定，通常為「豐富」；戰鬥類為「飽滿」。From the number of instruments and energy of the input songs; usually Rich, Full for battle styles.只有旋律、和聲、貝斯、鼓、襯底、填音 6 聲部（舊版效果）。Only lead, harmony, bass, drums, pad and fill (the previous sound).重複段加入高音琶音，B 段與高潮加入副旋律；各段力度漸強。Adds a high arpeggio in repeats and a counter-melody in the B section and climax; dynamics build by section.再加入旋律三度 / 六度和聲、輔助打擊樂（沙鈴、鈴鼓、三角鐵等）與高八度弦樂，層層堆疊。Also adds a third / sixth harmony under the lead, auxiliary percussion (shaker, tambourine, triangle...) and an upper string octave, layer by layer.所有層次從第一段起就加入，聲音最厚。All layers from the first section on; the thickest sound.不加填音。No fills.
No comments:
Post a Comment