Upload folder using huggingface_hub

.gitignore

@@ -0,0 +1,2 @@
+.DS_Store
+Thumbs.db

Code For Speech Generation/BSC/addNoise.py

@@ -0,0 +1,288 @@
+import numpy as np
+import soundfile as sf
+import os
+import glob
+from scipy import signal
+import random
+
+BIGCAT_TO_ENVS = {
+    "Domestic": ["DWASHING", "DKITCHEN", "DLIVING"],
+    "Nature": ["NFIELD", "NRIVER", "NPARK"],
+    "Office": ["OOFFICE", "OHALLWAY", "OMEETING"],
+    "Public": ["PSTATION", "PCAFETER", "PRESTO"],
+    "Street": ["STRAFFIC", "SPSQUARE", "SCAFE"],
+    "Transportation": ["TMETRO", "TBUS", "TCAR"],
+}
+
+ENV_TO_BIGCAT = {env: bigcat for bigcat, envs in BIGCAT_TO_ENVS.items() for env in envs}
+
+# Within each big category: rotate 3 sub-environment codes (Within-Mismatch)
+WITHIN_ROTATION = {
+    # Domestic
+    "DWASHING": "DKITCHEN",
+    "DKITCHEN": "DLIVING",
+    "DLIVING": "DWASHING",
+    # Nature
+    "NFIELD": "NRIVER",
+    "NRIVER": "NPARK",
+    "NPARK": "NFIELD",
+    # Office
+    "OOFFICE": "OHALLWAY",
+    "OHALLWAY": "OMEETING",
+    "OMEETING": "OOFFICE",
+    # Public
+    "PSTATION": "PCAFETER",
+    "PCAFETER": "PRESTO",
+    "PRESTO": "PSTATION",
+    # Street
+    "STRAFFIC": "SPSQUARE",
+    "SPSQUARE": "SCAFE",
+    "SCAFE": "STRAFFIC",
+    # Transportation
+    "TMETRO": "TBUS",
+    "TBUS": "TCAR",
+    "TCAR": "TMETRO",
+}
+
+# Cross big-category mapping (Cross-Mismatch)
+CROSS_BIGCAT_MAP = {
+    "Domestic": "Street",
+    "Nature": "Transportation",
+    "Office": "Nature",
+    "Public": "Domestic",
+    "Street": "Office",
+    "Transportation": "Public",
+}
+
+
+def read_wav(file_path):
+    """Read a WAV file."""
+    return sf.read(file_path)
+
+def write_wav(file_path, data, samplerate):
+    """Write a WAV file."""
+    sf.write(file_path, data, samplerate)
+
+def calculate_rms(signal):
+    """Compute RMS."""
+    return np.sqrt(np.mean(signal**2))
+
+def add_noise(speech, noise, snr_db):
+    """Add noise to speech at the given SNR (dB)."""
+    # Match speech and noise length
+    if len(noise) < len(speech):
+        # If noise is too short, tile it
+        noise = np.tile(noise, int(np.ceil(len(speech) / len(noise))))[:len(speech)]
+    else:
+        # If noise is too long, take a random segment
+        start = random.randint(0, len(noise) - len(speech))
+        noise = noise[start:start + len(speech)]
+    
+    # Compute RMS
+    speech_rms = calculate_rms(speech)
+    noise_rms = calculate_rms(noise)
+    
+    # Scale noise to target SNR
+    snr_linear = 10 ** (snr_db / 20)
+    noise_rms_target = speech_rms / snr_linear
+    
+    # Scale noise amplitude
+    noise_adjusted = noise * (noise_rms_target / (noise_rms + 1e-10))
+    
+    # Mix
+    noisy_speech = speech + noise_adjusted
+    
+    # Avoid clipping; normalize to [-1, 1]
+    max_val = np.max(np.abs(noisy_speech))
+    if max_val > 1:
+        noisy_speech = noisy_speech / max_val * 0.95
+    
+    return noisy_speech
+
+def parse_speech_stem(stem: str):
+    """
+    Parse speech filename (without extension).
+    Expected format:
+      - {ENV}_{E|I}{ID}   e.g. DKITCHEN_E01
+      - NEUTRAL_N{ID}     e.g. NEUTRAL_N03
+    Returns: (env_code, type_char, rest)
+    """
+    parts = stem.split("_", 1)
+    if len(parts) != 2:
+        raise ValueError(f"Cannot parse speech filename (missing '_'): {stem}")
+    env_code, rest = parts[0], parts[1]
+    if not rest:
+        raise ValueError(f"Cannot parse speech filename (empty after '_'): {stem}")
+    type_char = rest[0].upper()
+    return env_code.upper(), type_char, rest
+
+def build_noise_index(noise_files):
+    """Build noise env code -> file path mapping (stems uppercased)."""
+    idx = {}
+    for p in noise_files:
+        stem = os.path.splitext(os.path.basename(p))[0].upper()
+        idx[stem] = p
+    return idx
+
+def get_noise_path(noise_index, env_code: str):
+    """Look up noise file path by env code in the index (exact match after uppercasing)."""
+    env_code = env_code.upper()
+    return noise_index.get(env_code)
+
+def process_files(speech_folder, noise_folder, output_folder, snr_levels=[-10, -5, 0, 5 , 10], seed=None):
+    """Process all speech files."""
+    
+    # Collect all speech files
+    speech_files = glob.glob(os.path.join(speech_folder, "*.wav"))
+    speech_files.sort()
+    
+    # Collect all noise files (e.g. 06.wav, 07.wav, ...)
+    noise_files = glob.glob(os.path.join(noise_folder, "*.wav"))
+    noise_files.sort()
+    noise_index = build_noise_index(noise_files)
+
+    if seed is not None:
+        random.seed(seed)
+    
+    print(f"Found {len(speech_files)} speech files")
+    print(f"Found {len(noise_files)} noise files")
+    print(f"SNR levels (dB): {snr_levels}")
+    print("=" * 60)
+    
+    # Create output folder
+    os.makedirs(output_folder, exist_ok=True)
+    
+    # Count expected outputs:
+    # Explicit/Implicit: 3 conditions per utterance (Matched / Within / Cross); Neutral: 3 big-category backgrounds
+    total_files = 0
+    for speech_file in speech_files:
+        stem = os.path.splitext(os.path.basename(speech_file))[0]
+        try:
+            env_code, type_char, _ = parse_speech_stem(stem)
+        except Exception:
+            # Skip unparseable stems from the total count (they are skipped later too)
+            continue
+        if type_char == "N" or env_code == "NEUTRAL":
+            total_files += 3 * len(snr_levels)
+        else:
+            total_files += 3 * len(snr_levels)
+    processed = 0
+    
+    # Process each speech file
+    for speech_file in speech_files:
+        # Load speech
+        speech_data, speech_sr = read_wav(speech_file)
+        speech_name = os.path.splitext(os.path.basename(speech_file))[0]
+        try:
+            env_code, type_char, utt_id = parse_speech_stem(speech_name)
+        except Exception as e:
+            print(f"\nSkipping speech (unparseable filename): {speech_name}, reason: {e}")
+            continue
+        
+        print(f"\nProcessing speech: {speech_name} (duration: {len(speech_data)/speech_sr:.2f} s)")
+
+        # Pick 3 noise envs per utterance (Matched / Within / Cross, or Neutral rules)
+        selected_noise_envs = []
+
+        is_neutral = (type_char == "N") or (env_code == "NEUTRAL")
+        if is_neutral:
+            # Neutral: 3 different big-category backgrounds (no matched condition)
+            bigcats = random.sample(list(BIGCAT_TO_ENVS.keys()), 3)
+            for bigcat in bigcats:
+                noise_env = random.choice(BIGCAT_TO_ENVS[bigcat])
+                selected_noise_envs.append(noise_env)
+        else:
+            # Explicit / Implicit
+            if env_code not in ENV_TO_BIGCAT:
+                print(f"  Skip (unknown text env code): {env_code}")
+                continue
+
+            # 1) Matched
+            selected_noise_envs.append(env_code)
+
+            # 2) Within-Mismatch (rotate within same big category)
+            within_env = WITHIN_ROTATION.get(env_code)
+            if within_env is None:
+                print(f"  Warning: no Within rotation rule; skipping Within-Mismatch: {env_code}")
+            else:
+                selected_noise_envs.append(within_env)
+
+            # 3) Cross-Mismatch (cross big-category map + random sub-env in target category)
+            src_bigcat = ENV_TO_BIGCAT[env_code]
+            dst_bigcat = CROSS_BIGCAT_MAP.get(src_bigcat)
+            if dst_bigcat is None:
+                print(f"  Warning: no Cross big-category map; skipping Cross-Mismatch: {src_bigcat}")
+            else:
+                cross_env = random.choice(BIGCAT_TO_ENVS[dst_bigcat])
+                selected_noise_envs.append(cross_env)
+
+            # Require exactly 3 conditions when rules are complete
+            if len(selected_noise_envs) != 3:
+                print(f"  Skip (could not build 3 conditions, got {len(selected_noise_envs)}): {speech_name}")
+                continue
+
+        # Generate one output per selected noise env
+        for noise_env in selected_noise_envs:
+            noise_env = noise_env.upper()
+            noise_path = get_noise_path(noise_index, noise_env)
+            if noise_path is None:
+                print(f"  Skip (noise file not found): {noise_env}.wav")
+                continue
+
+            noise_data, noise_sr = read_wav(noise_path)
+            noise_name = os.path.splitext(os.path.basename(noise_path))[0].upper()
+
+            # Resample if sample rates differ
+            if speech_sr != noise_sr:
+                print(f"  Warning: sample rate mismatch - speech: {speech_sr} Hz, noise ({noise_name}): {noise_sr} Hz")
+                resample_ratio = speech_sr / noise_sr
+                new_length = int(len(noise_data) * resample_ratio)
+                noise_data = signal.resample(noise_data, new_length)
+
+            for snr in snr_levels:
+                noisy_speech = add_noise(speech_data, noise_data, snr)
+                # Output name: {speech_stem}_{noise_stem}_{snr}.wav
+                # e.g. DKITCHEN_E01_DKITCHEN_-5.wav
+                output_filename = f"{speech_name}_{noise_name}_{snr}.wav"
+                output_path = os.path.join(output_folder, output_filename)
+                write_wav(output_path, noisy_speech, speech_sr)
+                processed += 1
+
+        print(f"  Progress: {processed}/{total_files}")
+        print("-" * 40)
+    
+    print(f"\nDone. Generated {processed} files total.")
+    print(f"Saved under: {output_folder}")
+
+def check_files(folder):
+    """List a few WAV files in a folder."""
+    files = glob.glob(os.path.join(folder, "*.wav"))
+    print(f"\nFiles in folder {folder}:")
+    for f in files[:5]:  # show first 5 only
+        data, sr = read_wav(f)
+        print(f"  {os.path.basename(f)}: {sr} Hz, {len(data)/sr:.2f} s")
+    if len(files) > 5:
+        print(f"  ... and {len(files)-5} more file(s)")
+
+# Main
+if __name__ == "__main__":
+    # Folder paths
+    speech_folder = "speech_16k"  # speech WAVs
+    noise_folder = "noise"         # noise WAVs
+    output_folder = "noisy_speech" # output
+    
+    # SNR levels (dB)
+    snr_levels=[-10, -5, 0, 5 , 10]
+    
+    print("Starting speech + noise mixing...")
+    print(f"Speech folder: {speech_folder}")
+    print(f"Noise folder: {noise_folder}")
+    print(f"Output folder: {output_folder}")
+    
+    # Quick peek at inputs
+    check_files(speech_folder)
+    check_files(noise_folder)
+    
+    process_files(speech_folder, noise_folder, output_folder, snr_levels)
+    
+    print("\nAll done.")

Code For Speech Generation/BSC/edgeTTS.py

@@ -0,0 +1,76 @@
+import asyncio
+import edge_tts
+import pandas as pd
+import os
+from tqdm import tqdm  # progress bar
+
+# ================= Configuration =================
+INPUT_FILE = ""  
+OUTPUT_DIR = ""    
+VOICE = "en-US-GuyNeural"                                  
+# ===========================================
+
+async def generate_speech(code, text, output_path):
+    """
+    Generate one speech clip.
+    """
+    try:
+        communicate = edge_tts.Communicate(text, VOICE)
+        await communicate.save(output_path)
+        # Use tqdm.write instead of print so the progress bar layout stays intact
+        tqdm.write(f"[OK] Saved {code}.mp3")
+        return True
+    except Exception as e:
+        tqdm.write(f"[FAIL] {code}.mp3: {e}")
+        return False
+
+async def amain():
+    # 1. Ensure output directory exists
+    if not os.path.exists(OUTPUT_DIR):
+        os.makedirs(OUTPUT_DIR)
+        tqdm.write(f"Created output directory: {OUTPUT_DIR}")
+
+    # 2. Load Excel
+    try:
+        df = pd.read_excel(INPUT_FILE)
+        # Strip column names
+        df.columns = df.columns.str.strip()
+        if 'code' not in df.columns or 'sentence' not in df.columns:
+            tqdm.write("Error: Excel must contain 'code' and 'sentence' columns")
+            return
+    except FileNotFoundError:
+        tqdm.write(f"Error: file not found: {INPUT_FILE}")
+        return
+    except Exception as e:
+        tqdm.write(f"Error reading Excel: {e}")
+        return
+
+    total_count = len(df)
+    tqdm.write(f"Starting: {total_count} row(s)...")
+
+    # 3. Iterate rows with tqdm
+    # desc: label on the left of the bar
+    # unit: bar unit
+    # total: row count for percentage
+    for index, row in tqdm(df.iterrows(), total=total_count, desc="Progress", unit="row"):
+        code = str(row['code']).strip()
+        sentence = str(row['sentence']).strip()
+
+        # Skip empty sentences
+        if not sentence:
+            tqdm.write(f"[SKIP] row {index}: empty sentence")
+            continue
+
+        # Sanitize code for use in filenames
+        safe_code = "".join([c for c in code if c not in r'\/:*?"<>|'])
+        
+        # Output path
+        output_file = os.path.join(OUTPUT_DIR, f"{safe_code}.mp3")
+
+        # Synthesize
+        await generate_speech(safe_code, sentence, output_file)
+
+    tqdm.write("All tasks finished.")
+
+if __name__ == "__main__":
+    asyncio.run(amain())

Code For Speech Generation/BSC/mp3_to_wav.py

@@ -0,0 +1,68 @@
+from pydub import AudioSegment
+import os
+import glob
+
+def mp3_to_wav(mp3_path, output_folder=None, target_sample_rate=16000):
+    """Convert MP3 to WAV and set sample rate to 16 kHz."""
+
+    if output_folder:
+        filename = os.path.basename(mp3_path)
+        wav_filename = os.path.splitext(filename)[0] + '.wav'
+        wav_path = os.path.join(output_folder, wav_filename)
+    else:
+        wav_path = os.path.splitext(mp3_path)[0] + '.wav'
+
+    # Load MP3
+    audio = AudioSegment.from_mp3(mp3_path)
+    print(f"Original sample rate: {audio.frame_rate} Hz")
+
+    # Resample to target sample rate (16 kHz)
+    if audio.frame_rate != target_sample_rate:
+        audio = audio.set_frame_rate(target_sample_rate)
+        print(f"Resampled to: {target_sample_rate} Hz")
+
+    # Export as WAV
+    audio.export(wav_path, format="wav")
+
+    print(f"Conversion complete: {mp3_path} -> {wav_path}")
+    return wav_path
+
+def batch_convert(input_folder, output_folder=None, target_sample_rate=16000):
+
+    mp3_files = glob.glob(os.path.join(input_folder, "*.mp3"))
+    mp3_files.extend(glob.glob(os.path.join(input_folder, "*.MP3")))
+    mp3_files.sort()
+
+    if not mp3_files:
+        print(f"No MP3 files found in folder: {input_folder}")
+        return
+
+    if output_folder:
+        os.makedirs(output_folder, exist_ok=True)
+        print(f"Output folder: {output_folder}")
+
+    print(f"Found {len(mp3_files)} MP3 file(s), target sample rate: {target_sample_rate} Hz")
+    print("=" * 50)
+    success = 0
+    failed = 0
+
+    for mp3_file in mp3_files:
+        try:
+            mp3_to_wav(mp3_file, output_folder, target_sample_rate)
+            success += 1
+            print("-" * 50)
+        except Exception as e:
+            print(f"Conversion failed {mp3_file}: {e}")
+            failed += 1
+
+    print(f"Batch complete. Succeeded: {success}, failed: {failed}")
+
+if __name__ == "__main__":
+    input_folder = "speech"  # Input folder
+    output_folder = "speech_16k"  # Output folder for 16 kHz audio
+
+    if os.path.exists(input_folder):
+        batch_convert(input_folder, output_folder, target_sample_rate=16000)
+    else:
+        print(f"Folder does not exist: {input_folder}")
+        print(f"Current directory: {os.getcwd()}")

Code For Speech Generation/README.md

@@ -0,0 +1,23 @@
+
+# Speech generation code (DEAF)
+
+Scripts used to build the **BSC** and **SIC** speech datasets for [DEAF: A Benchmark for Diagnostic Evaluation of Acoustic Faithfulness in Audio Language Models](https://arxiv.org/abs/2603.18048).
+
+## BSC pipeline
+
+1. **`BSC/edgeTTS.py`** — Synthesize speech from text with Edge TTS.
+2. **`BSC/mp3_to_wav.py`** — Convert synthesized speech and background audio (e.g. MP3) to **WAV** at **16 kHz**.
+3. **`BSC/addNoise.py`** — Mix speech with background noise to produce the final **BSC** samples.
+
+Run these steps in order from the repository root or adjust paths to match your local layout.
+
+## SIC pipeline
+
+- **`SIC/SIC_audio_generation.py`** — Generate speech from text for the **SIC** subset (single script end-to-end for this track). Speech synthesis uses the **ElevenLabs Python SDK**.
+
+**Reference (tooling):** ElevenLabs. 2024. *ElevenLabs Python SDK.* [https://github.com/elevenlabs/elevenlabs-python](https://github.com/elevenlabs/elevenlabs-python)
+
+## Requirements
+
+Install dependencies used by the scripts you run (for example `edge-tts`, `pydub`, `soundfile`, `numpy`, `scipy`, **`elevenlabs`** for the SIC pipeline, and any others imported in each file). Pin versions in your own `requirements.txt` if you publish this folder as a standalone project.
+

Code For Speech Generation/SIC/SIC_audio_generation.py

@@ -0,0 +1,225 @@
+import os
+import re
+import wave
+from dataclasses import dataclass
+
+from dotenv import load_dotenv
+from elevenlabs.client import ElevenLabs
+from openpyxl import load_workbook
+
+load_dotenv()
+
+
+ELEVENLABS_API_KEY = " " # Enter your ElevenLabs API key here or set it in the .env file as ELEVENLABS_API_KEY
+SIC_XLSX_PATH = " "  # Enter the path to your SIC_texts.xlsx file here
+OUTPUT_DIR = "SIC_clips" # Output directory for generated clips
+MODEL_ID = "eleven_multilingual_v2"
+OUTPUT_FORMAT = "pcm_16000"  
+SAMPLE_RATE = 16000
+N_CHANNELS = 1
+SAMPLE_WIDTH_BYTES = 2  # 16-bit PCM
+
+
+VOICE_IDS = {
+    "elderly_male": "Av4Fi2idMFuA8kTbVZgv",
+    "young_male": "1wzJ0Fr9SDexsF2IsKU4",
+    "elderly_female": "0rEo3eAjssGDUCXHYENf",
+    "young_female": "aFueGIISJUmscc05ZNfD",
+}
+
+# For single-dimension contrasts, to control only that dimension,
+# the other dimension is fixed by default:
+# - Gender dimension: fix age to "young"
+# - Age dimension: fix gender to "male"
+GDR_FIXED_AGE = "young"   # "young" or "elderly"
+AGE_FIXED_GENDER = "male" # "male" or "female"
+
+
+@dataclass(frozen=True)
+class SicCode:
+    dim: str    # GDR / AGE / CMB / NEU
+    typ: str    # EX / IM / NT
+    ident: str  # F/M/EL/YG/EF/EM/YF/YM/NA
+    nn: str     # 01,02,...
+
+    @property
+    def stem(self) -> str:
+        return f"{self.dim}_{self.typ}_{self.ident}_{self.nn}"
+
+
+CODE_RE = re.compile(r"^(GDR|AGE|CMB|NEU)_(EX|IM|NT)_([A-Z]{1,2})_(\d{2})$")
+
+
+def parse_code(code: str) -> SicCode:
+    code = str(code).strip().upper()
+    m = CODE_RE.match(code)
+    if not m:
+        raise ValueError(f"Invalid code format: {code}")
+    return SicCode(dim=m.group(1), typ=m.group(2), ident=m.group(3), nn=m.group(4))
+
+
+def voice_key_from_gender_age(gender: str, age: str) -> str:
+    gender = gender.lower()
+    age = age.lower()
+    if gender not in ("male", "female"):
+        raise ValueError(f"Unknown gender: {gender}")
+    if age not in ("young", "elderly"):
+        raise ValueError(f"Unknown age: {age}")
+    return f"{age}_{gender}"
+
+
+def pick_voices_for_code(sc: SicCode) -> list[str]:
+    """
+    Return the list of `voice_key`s to generate (each key must exist in `VOICE_IDS`).
+    - GDR: 2 clips (matched / mismatched)
+    - AGE: 2 clips (matched / mismatched)
+    - CMB: 4 clips (both matched / gender mismatched only / age mismatched only / both mismatched)
+    - NEU or NT: 4 clips (young_male, young_female, elderly_male, elderly_female)
+    """
+    # Neutral (control group)
+    if sc.dim == "NEU" or sc.typ == "NT":
+        return ["young_male", "young_female", "elderly_male", "elderly_female"]
+
+    if sc.dim == "GDR":
+        if sc.ident not in ("F", "M"):
+            raise ValueError(f"GDR ident must be F/M, got: {sc.ident}")
+        gender_sem = "female" if sc.ident == "F" else "male"
+        age_fixed = GDR_FIXED_AGE
+        matched = voice_key_from_gender_age(gender_sem, age_fixed)
+        mismatched_gender = "male" if gender_sem == "female" else "female"
+        mismatched = voice_key_from_gender_age(mismatched_gender, age_fixed)
+        return [matched, mismatched]
+
+    if sc.dim == "AGE":
+        if sc.ident not in ("EL", "YG"):
+            raise ValueError(f"AGE ident must be EL/YG, got: {sc.ident}")
+        age_sem = "elderly" if sc.ident == "EL" else "young"
+        gender_fixed = AGE_FIXED_GENDER
+        matched = voice_key_from_gender_age(gender_fixed, age_sem)
+        mismatched_age = "young" if age_sem == "elderly" else "elderly"
+        mismatched = voice_key_from_gender_age(gender_fixed, mismatched_age)
+        return [matched, mismatched]
+
+    if sc.dim == "CMB":
+        if sc.ident not in ("EF", "EM", "YF", "YM"):
+            raise ValueError(f"CMB ident must be EF/EM/YF/YM, got: {sc.ident}")
+        age_sem = "elderly" if sc.ident.startswith("E") else "young"
+        gender_sem = "female" if sc.ident.endswith("F") else "male"
+
+        both_matched = voice_key_from_gender_age(gender_sem, age_sem)
+
+        # gender mismatched only: flip gender keep age
+        g_flip = "male" if gender_sem == "female" else "female"
+        gender_mis_only = voice_key_from_gender_age(g_flip, age_sem)
+
+        # age mismatched only: flip age keep gender
+        a_flip = "young" if age_sem == "elderly" else "elderly"
+        age_mis_only = voice_key_from_gender_age(gender_sem, a_flip)
+
+        # both mismatched: flip both
+        both_mis = voice_key_from_gender_age(g_flip, a_flip)
+
+        return [both_matched, gender_mis_only, age_mis_only, both_mis]
+
+    raise ValueError(f"Unknown dim: {sc.dim}")
+
+
+def pcm16_to_wav_bytes(pcm_bytes: bytes) -> bytes:
+    """Wrap ElevenLabs `pcm_16000` (16-bit) into WAV bytes."""
+    out_path = None
+    # Write WAV via `wave`: use a temp file to avoid extra dependencies (keep it simple and stable)
+    import tempfile
+    with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as tmp:
+        out_path = tmp.name
+    try:
+        with wave.open(out_path, "wb") as wf:
+            wf.setnchannels(N_CHANNELS)
+            wf.setsampwidth(SAMPLE_WIDTH_BYTES)
+            wf.setframerate(SAMPLE_RATE)
+            wf.writeframes(pcm_bytes)
+        with open(out_path, "rb") as f:
+            return f.read()
+    finally:
+        if out_path and os.path.exists(out_path):
+            try:
+                os.remove(out_path)
+            except OSError:
+                pass
+
+
+def generate_one(client: ElevenLabs, text: str, voice_key: str, out_path: str):
+    voice_id = VOICE_IDS.get(voice_key)
+    if not voice_id:
+        raise ValueError(f"voice_key not configured in VOICE_IDS: {voice_key}")
+
+    audio_stream = client.text_to_speech.convert(
+        text=text,
+        voice_id=voice_id,
+        model_id=MODEL_ID,
+        output_format=OUTPUT_FORMAT,
+    )
+
+    pcm = b"".join(audio_stream)
+    wav_bytes = pcm16_to_wav_bytes(pcm)
+    with open(out_path, "wb") as f:
+        f.write(wav_bytes)
+
+
+def load_sic_rows(xlsx_path: str) -> list[tuple[str, str]]:
+    wb = load_workbook(xlsx_path)
+    ws = wb.active
+    rows = []
+    for r in ws.iter_rows(min_row=1, values_only=True):
+        if not r or len(r) < 2:
+            continue
+        code, sentence = r[0], r[1]
+        if code is None or sentence is None:
+            continue
+        code_s = str(code).strip()
+        sent_s = str(sentence).strip()
+        if not code_s or code_s.lower() == "code":
+            continue
+        rows.append((code_s, sent_s))
+    return rows
+
+
+def main():
+    os.makedirs(OUTPUT_DIR, exist_ok=True)
+    client = ElevenLabs(api_key=ELEVENLABS_API_KEY)
+
+    rows = load_sic_rows(SIC_XLSX_PATH)
+    print(f"Loaded {len(rows)} rows")
+
+
+    total_clips = 0
+    for code_raw, _ in rows:
+        try:
+            sc_tmp = parse_code(code_raw)
+            total_clips += len(pick_voices_for_code(sc_tmp))
+        except Exception:
+            continue
+
+    ok = 0
+    fail = 0
+    for code_raw, sentence in rows:
+        try:
+            sc = parse_code(code_raw)
+            voice_keys = pick_voices_for_code(sc)
+            for vk in voice_keys:
+                out_name = f"{sc.stem}__{vk}.wav"
+                out_path = os.path.join(OUTPUT_DIR, out_name)
+                generate_one(client, sentence, vk, out_path)
+                ok += 1
+                if total_clips:
+                    progress = ok + fail
+                    pct = progress * 100.0 / total_clips
+                    print(f"[progress] {progress}/{total_clips} ({pct:.1f}%) - current: {out_name}")
+        except Exception as e:
+            fail += 1
+            print(f"[skip] {code_raw}: {e}")
+
+    print(f"Done: generated {ok} clips; skipped/failed {fail} rows. Output directory: {OUTPUT_DIR}")
+
+
+if __name__ == "__main__":
+    main()

Data/Audio/BSC/DKITCHEN_E01.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dac697aacfff0491148e8c56e43d48b172c566f83011522aac8aa9ffff9d99fa
+size 172844

Data/Audio/BSC/DKITCHEN_E02.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:44a5a10bb1166db29585560ddeb6a331b87da553364998ee0900ccfdb03bc954
+size 211244

Data/Audio/BSC/DKITCHEN_I01.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7a57d1390c6d3e34d8a5c24658fd4a89e525130f591bf05fddab7be05081480f
+size 146732

Data/Audio/BSC/DKITCHEN_I02.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f13412e0368a6a489dcf46df34e21446f9d2f5dd19bec7258ba24a9f84f4a86c
+size 302636

Data/Audio/BSC/DLIVING_E01.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ce62eb015d8a0109bd4b34f01b3420a0c4a653349512a4cb74a126ef9c78e79b
+size 159788

Data/Audio/BSC/DLIVING_E02.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ede90ef9878a71000c8160e7372051e411d0fa4704d7a5ae8dce3fcf58583c3f
+size 215084

Data/Audio/BSC/DLIVING_I01.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6aab865e3213c62cd95331f1c710e35bdb09f8d461b35928f404edf2abc3c02c
+size 192044

Data/Audio/BSC/DLIVING_I02.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2779f383758421fbd43df613e9464e359fb960274744ae06c66b63dd146bd7b0
+size 195116

Data/Audio/BSC/DWASHING_E01.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b0d365ccdfb2ef2613917e4b9ae19d9f15be461d334ff8c8b17f277644879117
+size 233516

Data/Audio/BSC/DWASHING_E02.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:25e7d4c00430d35a796f42add63d5de1dff733e963674c9c2870b4defd4b5ba8
+size 191276

Data/Audio/BSC/DWASHING_I01.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ca582c436df4479b1ecf8fa4a965c27a15152963207cebb52346018b2fe08e0f
+size 248108

Data/Audio/BSC/DWASHING_I02.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6a23c3c0f98ca2c6a3c33fc8a8c4781288de84533e299f10db0f4b973e477992
+size 214316

Data/Audio/BSC/NEUTRAL_N01.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:90bfb10516c4631c352accca6b12f0d5abbeaab662857c497c33f18a9f468342
+size 107564

Data/Audio/BSC/NEUTRAL_N02.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cdbf0cbd5a69f2c5332238837be93b92a18f1c33b619e0f8f4df9c079e30a2b6
+size 131372

Data/Audio/BSC/NEUTRAL_N03.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a61c52691d8bc8e4ad97c18c69b53c18ad744165e68f9ad7494901d7ee810969
+size 119084

Data/Audio/BSC/NEUTRAL_N04.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a3b1810e4711b68b4e7a53d60440221d42dc780dc065a42fd44e13edd390f401
+size 124460

Data/Audio/BSC/NEUTRAL_N05.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:02149d4b784806abc9e12a98ec2984c977d36d36ef67e879eb493e501d7b025d
+size 119084

Data/Audio/BSC/NEUTRAL_N06.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:38d1521c94bd6dda708930692c15992ae4f90866433881311166b35051a55394
+size 140588

Data/Audio/BSC/NEUTRAL_N07.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7591cacecd6f87dfd9db2142e70f2c0b8e763619035129bfe39f6d1479e88239
+size 132140

Data/Audio/BSC/NEUTRAL_N08.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8b6cac4d2a4e10c729fabec8f23f7188bc9128c7ee6d4b6ab7248855e2bf44cd
+size 156716

Data/Audio/BSC/NEUTRAL_N09.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:165d274215b43958b27d6aee1bf4b8e499776f116ef0a20f4e56cd1914717688
+size 103724

Data/Audio/BSC/NEUTRAL_N10.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d93003adff76c18dce60cee6bd0dfb485743105662e8a1e348d3947e2ca26763
+size 129068

Data/Audio/BSC/NEUTRAL_N11.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:36cbedc29a5735df930c0dfc27c8b0a529fbb251f82ce105dde3622fd77a484e
+size 114476

Data/Audio/BSC/NEUTRAL_N12.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3a099eda3008adaddba59f22ffcf9377edfdbd45e814fdb6d6ea561cc87240ae
+size 129836

Data/Audio/BSC/NFIELD_E01.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9eb927ba20e6958e17bee07cc9c5a9ec3e45872961351267d18a12ac3844c982
+size 143660

Data/Audio/BSC/NFIELD_E02.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2a42ae48f343208c5850a656af1a5bf747e011b3cffa1328f6f28a8423bf745e
+size 258092

Data/Audio/BSC/NFIELD_I01.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d395a0f5ad6d70a53a7fa8ab7b698070c5f4b52908ed73d1ca9e2f819decc759
+size 249644

Data/Audio/BSC/NFIELD_I02.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b8ac26774e61e90badcab0bdae47075935af9d0b3a60548acf2878d9cacaeb17
+size 247340

Data/Audio/BSC/NPARK_E01.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ea7f61c621f7a4207920d0578f57d9a188c784af1616cff0ff96796f84d59664
+size 258860

Data/Audio/BSC/NPARK_E02.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3f47253bbc7ae6c864b531a00a9a94390901b5016c1fae921ea8d1c3c9d04aee
+size 284204

Data/Audio/BSC/NPARK_I01.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f64f7d8ae5c6d59b69a7ac1b495ead8b487e63529057736f1045ecc5c1cc1fff
+size 188972

Data/Audio/BSC/NPARK_I02.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:69f18734a66434cb93440101c299e2bb41cf2945369f9542d1df0a5520ee8073
+size 234284

Data/Audio/BSC/NRIVER_E01.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7d57e94e4b818f8ba03295ec9919116d79a2412b7e5dcf74963a3c90087574b0
+size 183596

Data/Audio/BSC/NRIVER_E02.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:077698ec1903e2006359305488105b0ef84e4ae03e4a113f84aebd0cd6fff757
+size 174380

Data/Audio/BSC/NRIVER_I01.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:313d3c64b55ddb40fc7f606f823bb788be7bf1d0e8212d07de3eb01972d6e3a9
+size 178220

Data/Audio/BSC/NRIVER_I02.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:18cdfba2b3deff930e4883d68f368d30269570225fc59531d7ffd9a20aa1f9cc
+size 196652

Data/Audio/BSC/OHALLWAY_E01.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8f6b1a8c2992ed6f547687e6d9c72fc661b0dd916465e09adea7b35b0e2b7140
+size 181292

Data/Audio/BSC/OHALLWAY_E02.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1097d4e65461e48aee0e6b6a5fef029b3a35c8d8a34ed251f964500515555365
+size 155948

Data/Audio/BSC/OHALLWAY_I01.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c3742119bf6fcecf10105f37558a7d095f630b6af180d1362bef58ef46d534d4
+size 176684

Data/Audio/BSC/OHALLWAY_I02.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:abffdb050a09256b81f7c81c3a6704d79924b7dc58a06febe5253944530ae9ff
+size 188972

Data/Audio/BSC/OMEETING_E01.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a1ffd0bea9e8d635cc5e390e26aa70f372abdf72f646350fd93a5f82473fe253
+size 225068

Data/Audio/BSC/OMEETING_E02.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1f6dbbfb38cc3b3dd06666b29f2c1b01de0185e9ba2d2d5408955d7ba5ad073e
+size 202796

Data/Audio/BSC/OMEETING_I01.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9745fc0ea81251e46de24533aaaf8c47bd471950bc9c64387a75adac9b6cd1ff
+size 220460

Data/Audio/BSC/OMEETING_I02.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:58776c9835602d72ba6e97465335874876319a882271a03f81f529059503a741
+size 177452