Text-to-Speech API Error - Empty Input String Issue

Other Questions OpenAI
1

I’m building an application that processes data and converts it to speech audio files using OpenAI’s TTS API. However, I keep getting a BadRequestError saying the input string is too short.

Here’s my voice generator module:

import openai

def process_content(text_data):
    results = []
    segments = text_data.split("\n")
    for segment in segments:
        if segment.startswith('Scene: '):
            scene_desc = segment.replace("Scene:", "Scene: ")
            results.append({
                "category": "visual",
                "info": scene_desc
            })
        elif segment.startswith("Voice:"):
            speech_text = segment.replace("Voice:", "Voice:")
            results.append({
                "category": "audio",
                "text": speech_text
            })
    return results

def generate_audio(content_list, file_path="final_audio.mp3"):
    speech_content = ""
    for item in content_list:
        if item["category"] != 'audio':
            continue
        speech_content += item["text"] + "\n\n"
    
    voice_output = openai.audio.speech.create(
        model="tts-1",
        input=speech_content,
        voice="nova"
    )
    voice_output.stream_to_file(file_path)

And my main script:

with open("input_data.txt", "r") as file:
    raw_data = file.read()

ai_response = openai.chat.completions.create(
    model="gpt-3.5-turbo",
    messages=[
        {
            "role": "system",
            "content": """Generate content for a short presentation..."""
        },
        {
            "role": "user", 
            "content": f"Process this data: {raw_data}"
        }
    ]
)

processed_data = voice_gen.process_content(ai_response.choices[0].message.content)
voice_gen.generate_audio(processed_data, "presentation.mp3")

The error shows that the input parameter is empty when calling the speech API. I think my text parsing logic might be filtering out all the content, leaving nothing for the TTS function. Has anyone encountered this before?

2

Had the same issue with OpenAI’s TTS API last month. Your text extraction logic is the problem - you’re replacing “Voice:” with “Voice:” which doesn’t actually remove anything. Try speech_text = segment.replace("Voice:", "").strip() to strip out the prefix and get clean text. I’d add a quick validation like if not speech_content.strip(): before hitting the API to catch empty inputs. Also heads up - GPT responses don’t always match your expected format exactly, so maybe use partial string matching instead of looking for exact prefixes.

3

your parsing logic’s not quite right - in segment.replace("Voice:", "Voice:"), you’re not removing anything. try changing it to speech_text = segment.replace("Voice:", "").strip(). also, add some debug prints to see if speech_content is empty before calling the API.

4

yeah, that’s your problem - you’re replacing “Voice:” with “Voice:” so nothing changes. also, check if speech_content is empty before hitting openAI. try adding print(repr(speech_content)) to see what you’re actually getting.