ComfyUI Node: 🔊➜📝 STT - Speech to Text

🔊➜📝 STT - Speech to Text:

The Bjornulf_SpeechToText node is designed to convert spoken language into written text, a process known as speech-to-text transcription. This node is particularly useful for AI artists and developers who need to transcribe audio content into text format for further processing or analysis. It supports various audio input methods, including direct audio data and file paths, making it versatile for different use cases. The node is capable of detecting the language of the spoken content, which enhances its utility in multilingual environments. By leveraging local transcription models, it ensures that the transcription process is efficient and can be tailored to different levels of accuracy and performance based on the model size selected. This node is essential for applications that require converting audio inputs into text, such as creating subtitles, transcribing interviews, or processing voice commands.

🔊➜📝 STT - Speech to Text Input Parameters:

model_size

The model_size parameter determines the size of the transcription model used for processing the audio input. It impacts the accuracy and speed of the transcription process. Available options are "tiny", "base", "small", "medium", and "large-v2", with "base" being the default. Smaller models are faster but may be less accurate, while larger models provide better accuracy at the cost of increased processing time.

AUDIO

The AUDIO parameter is an optional input that allows you to provide audio data directly in tensor format. This is useful when the audio is already available in a digital format and needs to be processed without saving it to a file. The parameter expects a dictionary containing waveform and sample_rate keys, which represent the audio data and its sampling rate, respectively.

audio_path

The audio_path parameter is an optional string input that specifies the file path to an audio file that needs to be transcribed. This parameter is useful when the audio content is stored in a file and needs to be processed. It is important to ensure that the file path is valid and accessible to avoid errors during transcription.

🔊➜📝 STT - Speech to Text Output Parameters:

transcript

The transcript output provides the transcribed text from the audio input. It represents the spoken content in written form, allowing you to use it for further text-based processing or analysis. If the transcription process fails, this output will contain an error message indicating the failure.

detected_language

The detected_language output indicates the language code of the spoken content detected during the transcription process. This information is useful for understanding the language context of the audio input and can be used to tailor subsequent processing steps accordingly.

language_name

The language_name output provides the full name of the detected language, making it easier to interpret the language code provided by the detected_language output. This output enhances the readability and usability of the transcription results, especially in multilingual applications.

🔊➜📝 STT - Speech to Text Usage Tips:

• Ensure that the audio input is clear and free from background noise to improve transcription accuracy.
• Choose the appropriate model_size based on your accuracy and performance needs; larger models offer better accuracy but require more processing time.
• Verify that the audio_path is correct and accessible to avoid errors during the transcription process.

🔊➜📝 STT - Speech to Text Common Errors and Solutions:

No valid audio input provided

• Explanation: This error occurs when neither the AUDIO parameter nor the audio_path is provided or valid.
• Solution: Ensure that you provide a valid audio input either through the AUDIO parameter or by specifying a correct audio_path.

Local transcription failed: <error_message>

• Explanation: This error indicates that the local transcription process encountered an issue, with <error_message> providing specific details.
• Solution: Check the error message for details and ensure that the audio input is in a supported format and that the model is correctly configured.