text to speech (tts) .docx script for python using kokoro

gistfile1.txt

# tts_from_docx.py

# make sure in terminal this command has run and kokoro is installed: pip install kokoro>=0.9.2 soundfile python-docx

from kokoro import KPipeline
import soundfile as sf
from docx import Document
import numpy as np
import os


# CONFIGURATION
##########################################

DOCX_FILE = "abc.docx"              # Path to your input file
OUTPUT_DIR = "audio_parts"          # Folder for chunks
FINAL_AUDIO = "audio.wav"           # Final combined file
LANG_CODE = 'a'                     # 'a' = auto
VOICE = 'en_heart'                  # Kokoro voice en_heart = standard
CHUNK_SIZE = 500                    # Characters per chunk
SAMPLE_RATE = 24000                 # Audio sample rate


# 1. Load DOCX
##########################################

def extract_text_from_docx(docx_path):
    doc = Document(docx_path)
    text = "\n".join(p.text for p in doc.paragraphs if p.text.strip())
    return text

print(f"Reading DOCX: {DOCX_FILE}")
full_text = extract_text_from_docx(DOCX_FILE)
print(f"Total text length: {len(full_text)} characters")


# 2. Split text into chunks
##########################################

def split_text(text, max_chars=CHUNK_SIZE):
    paragraphs = text.split("\n")
    chunks = []
    chunk = ""
    for p in paragraphs:
        if len(chunk) + len(p) < max_chars:
            chunk += p + "\n"
        else:
            chunks.append(chunk.strip())
            chunk = p + "\n"
    if chunk.strip():
        chunks.append(chunk.strip())
    return chunks

chunks = split_text(full_text)
print(f"Split into {len(chunks)} chunks of ~{CHUNK_SIZE} characters")


# 3. Create output folder
##########################################

if not os.path.exists(OUTPUT_DIR):
    os.makedirs(OUTPUT_DIR)


# 4. Initialize Kokoro TTS
##########################################

print("Loading Kokoro pipeline...")
pipeline = KPipeline(lang_code=LANG_CODE)


# 5. Generate audio for each chunk
##########################################

all_audio_files = []

for i, chunk in enumerate(chunks):
    print(f"Generating audio for chunk {i+1}/{len(chunks)}")
    generator = pipeline(chunk, voice=VOICE)
    for j, (gs, ps, audio) in enumerate(generator):
        filename = os.path.join(OUTPUT_DIR, f"part_{i}_{j}.wav")
        sf.write(filename, audio, SAMPLE_RATE)
        all_audio_files.append(filename)
        print(f"Saved {filename}")


# 6. Combine all chunks into single WAV
##########################################

print("Combining audio files into one audiobook...")
combined_audio = []

for filename in all_audio_files:
    audio_data, sr = sf.read(filename)
    if sr != SAMPLE_RATE:
        raise ValueError(f"Sample rate mismatch in {filename}")
    combined_audio.append(audio_data)

final_audio_data = np.concatenate(combined_audio)
sf.write(FINAL_AUDIO, final_audio_data, SAMPLE_RATE)

print(f"🎧 Audiobook created: {FINAL_AUDIO}")


# 7. Delete individual chunk files
##########################################

print("🧹 Cleaning up individual audio chunks...")
for filename in all_audio_files:
    try:
        os.remove(filename)
        print(f"🗑️ Deleted {filename}")
    except Exception as e:
        print(f"⚠️ Could not delete {filename}: {e}")

print("✅ All done! Created by SharedAcademia.org")