ASR_Server.py

from flask import Flask, jsonify
from datasets import load_dataset, Audio
import pandas as pd
import os
import threading

import os
os.environ["HF_HOME"] = "/tmp/huggingface"


csv_path = "test.csv" 
df = pd.read_csv(csv_path)
print(f"CSV Loaded with {len(df)} rows")

# def generateTranscript(ASR_model, dataset, csv_path, output_dir="./"):
#     import os
#     import time
#     import pandas as pd
#     import librosa
#     import tqdm
#     from transformers import pipeline

#     os.makedirs(output_dir, exist_ok=True)
#     # output_csv_path = os.path.join(output_dir, f"test_with_{ASR_model.replace('/', '_')}.csv")
#     output_csv_path = os.path.join(output_dir, f"test_with_{ASR_model}.csv")

#     if os.path.exists(output_csv_path):
#         print(f"Transcript already exists for model {ASR_model}. Skipping transcription.")
#         return

#     # Load metadata CSV
#     df = pd.read_csv(csv_path)
#     print(f"CSV Loaded with {len(df)} rows")

#     # Prepare
#     df[df.columns[0]] = df[df.columns[0]].str.strip().str.lower()
#     filename_column = df.columns[0]
#     transcripts = []
#     rtfx_score = []

#     # Load ASR model
#     pipe = pipeline("automatic-speech-recognition", model=ASR_model)

#     # Create a map of dataset samples by file name (assumes filename is in dataset)
#     dataset_map = {
#         sample["audio"]["path"].split("/")[-1].lower(): sample for sample in dataset
#     }

#     for idx, row in tqdm.tqdm(df.iterrows(), total=len(df)):
#         filename = row[filename_column].strip().lower() + ".wav"
#         if filename in dataset_map:
#             sample = dataset_map[filename]
#             try:
#                 audio_array = sample["audio"]["array"]
#                 sample_rate = sample["audio"]["sampling_rate"]

#                 start_time = time.time()
#                 result = pipe({"array": audio_array, "sampling_rate": sample_rate})
#                 end_time = time.time()

#                 transcript = result["text"]
#                 duration = librosa.get_duration(y=audio_array, sr=sample_rate)
#                 rtfx = (end_time - start_time) / duration if duration > 0 else 0

#                 transcripts.append(transcript)
#                 rtfx_score.append(rtfx)

#                 print(f"✅ {filename}: RTFX = {rtfx:.2f}")

#             except Exception as e:
#                 print(f"❌ Error with {filename}: {e}")
#                 transcripts.append("")
#                 rtfx_score.append(0)
#         else:
#             print(f"⚠️ File not in dataset: {filename}")
#             transcripts.append("")
#             rtfx_score.append(0)

#     # Append to original DataFrame
#     df['transcript'] = transcripts
#     df['rtfx'] = rtfx_score

#     df.to_csv(output_csv_path, index=False)
#     print(f"✅ Transcripts saved to {output_csv_path}")

def generateTranscript(ASR_model, csv_path, output_dir="./"):
    import os
    import time
    import tqdm
    import pandas as pd
    import soundfile as sf
    from transformers import pipeline

    # Load dataset without decoding audio (required!)
    dataset = load_dataset("satyamr196/asr_fairness_audio", split="train")
    # dataset = dataset.with_format("python", decode_audio=False)
    dataset = dataset.cast_column("audio", Audio(decode=False))

    output_csv_path = os.path.join("./", f"test_with_{ASR_model}.csv")
    # Check if transcript already exists
    if os.path.exists(output_csv_path):
        print(f"Transcript already exists for model {ASR_model}. Skipping transcription.")
        return

    # Load CSV
    df = pd.read_csv(csv_path)
    print(f"CSV Loaded with {len(df)} rows")

    # Initialize ASR pipeline
    pipe = pipeline("automatic-speech-recognition", model=ASR_model, device=-1)
    # print("Device set to use CPU")

    # Column with filenames in the CSV
    filename_column = df.columns[0]
    df[filename_column] = df[filename_column].str.strip().str.lower()

    # Build map from filename -> dataset sample (without decoding audio)
    # print("Creating dataset map from filenames...")
    # dataset = dataset.with_format("python", decode_audio=False)
    dataset_map = {
        os.path.basename(sample["audio"]["path"]).lower(): sample
        for sample in dataset
    }

    transcripts = []
    rtfx_score = []

    for idx, row in tqdm.tqdm(df.iterrows(), total=len(df)):
        filename = row[filename_column] + ".wav"

        if filename in dataset_map:
            sample = dataset_map[filename]
            try:
                # Decode audio only when needed
                file_path = sample["audio"]["path"]
                audio_array, sample_rate = sf.read(file_path)

                start_time = time.time()
                result = pipe({"array": audio_array, "sampling_rate": sample_rate})
                end_time = time.time()

                transcript = result["text"]
                duration = len(audio_array) / sample_rate
                rtfx = (end_time - start_time) / duration if duration > 0 else 0

                transcripts.append(transcript)
                rtfx_score.append(rtfx)

                print(f"✅ {filename}: RTFX = {rtfx:.2f}")

            except Exception as e:
                print(f"❌ Error with {filename}: {e}")
                transcripts.append("")
                rtfx_score.append(0)
        else:
            print(f"❌ File not found in dataset: {filename}")
            transcripts.append("")
            rtfx_score.append(0)

    # Save results
    df["transcript"] = transcripts
    df["rtfx"] = rtfx_score

    os.makedirs(output_dir, exist_ok=True)
    # Create the directory if it doesn't exist
    output_dir = os.path.dirname(os.path.join(output_dir, f"test_with_{ASR_model}.csv")) # Get the directory path
    if not os.path.exists(output_dir):  # Check if directory exists
        os.makedirs(output_dir)  # Create directory if it doesn't exist
        print(f"Created directory: {output_dir}")
        
    df.to_csv(output_csv_path, index=False)

    print(f"\n📄 Transcripts saved to: {output_csv_path}")



app = Flask(__name__)

@app.route("/")
def home():
    return jsonify(
        {
            "message": "Welcome to the ASR Server! Please use the App link to access ASR-FairBench application.",
            "App link": "https://github.com/SatyamR196/ASR-FairBench"
        }
    )

@app.route("/asr_models")
def asr_models():
    models = [
        "DeepSpeech",
        "Wav2Vec",
        "Jasper",
        "QuartzNet",
        "Conformer",
        "whisper",
        "Kaldi",
        "SpeechBrain",
        "Fairseq S2T",
        "ESPnet"
    ]
    def background_job():
        generateTranscript("openai/whisper-base", csv_path, output_dir="./")

    # Start the background job in a separate thread
    threading.Thread(target=background_job).start()
    print("Transcription started in background")
    return jsonify({"asr_models": models})

# if __name__ == "__main__":
#     app.run(debug=True)