Upload 4 files

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+soundfont.sf2 filter=lfs diff=lfs merge=lfs -text

app.py

@@ -1,17 +1,46 @@
 import gradio as gr
+import os
 from infer import infer_midi_from_wav
+import subprocess
 
-def transcribe(wav_path):
-    midi_path = infer_midi_from_wav(wav_path)
-    return midi_path
+BASE_DIR = os.path.dirname(os.path.abspath(__file__))
 
-demo = gr.Interface(
-    fn=transcribe,
-    inputs=gr.Audio(type="filepath"),  # ✅ 修正ポイント
-    outputs=gr.File(label="Download MIDI"),
-    title="Humming to MIDI",
-    description="Upload your humming (WAV) to get MIDI output"
+# MIDI → WAV 変換関数（fluidsynth使用）
+def convert_midi_to_wav(midi_path):
+    soundfont_path = os.path.join(BASE_DIR, "soundfont.sf2")  # SoundFontファイルを用意
+    wav_path = os.path.join(BASE_DIR, "synth_output.wav")
+
+    command = [
+        "fluidsynth",
+        "-ni",
+        soundfont_path,
+        midi_path,
+        "-F",
+        wav_path,
+        "-r",
+        "44100"
+    ]
+    result = subprocess.run(command, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
+    if result.returncode != 0:
+        raise RuntimeError("fluidsynth conversion failed:\n" + result.stderr.decode())
+    return wav_path
+
+# Gradioで呼び出す推論＆再生用関数
+def transcribe_and_play(wav_file):
+    midi_path = infer_midi_from_wav(wav_file)
+    wav_output_path = convert_midi_to_wav(midi_path)
+    return wav_output_path, midi_path
+
+# Gradio UI
+interface = gr.Interface(
+    fn=transcribe_and_play,
+    inputs=gr.Audio(source="microphone", type="filepath", label="\u97f3\u58f0\u9332\u97f3"),
+    outputs=[
+        gr.Audio(label="\u30d4\u30a2\u30ce\u97f3\u3067\u518d\u751f"),
+        gr.File(label="MIDI\u30c0\u30a6\u30f3\u30ed\u30fc\u30c9")
+    ],
+    title="\u9f3b\u6b4c\u304b\u3089\u306eMIDI\u5909\u63db\u30c7\u30e2",
+    description="\u9332\u97f3\u3057\u305f\u97f3\u58f0\u3092MIDI\u306b\u5909\u63db\u3057\u3001\u30d4\u30a2\u30ce\u97f3\u3067\u518d\u751f\u3057\u307e\u3059"
 )
 
-if __name__ == "__main__":
-    demo.launch()
+interface.launch()

app.txt

@@ -0,0 +1 @@
+fluidsynth

requirements.txt

@@ -17,6 +17,8 @@ tqdm
 librosa
 soundfile
 gradio
+fluidsynth
+
 
 
 # MIDI・音楽処理

soundfont.sf2

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:74594e8f4250680adf590507a306655a299935343583256f3b722c48a1bc1cb0
+size 148398306