Free temporary sound transcription (speech in the text)?

Question

Is there a free service / program that will transcribe audio into text and save timestamps?

I want to decrypt:

This is a demo link for VoiceBase

(A 16-minute conversation between 419 scammers, scammers speak with heavy accents, so I do not complain about the quality of transcription).

Voicebase records it, but transcription is one block of the test and has no time limits.

I also tried the “trick” of Google’s voice (write the file to voicemail and Google transcribes it for 3 minutes at a time), but, again, Transcription is a single block of text, not timestampped.

I know the Sphinx, but could not get it to work.

I understand that I could use subtitleeditor to create a manual trascription, but I'm looking for an automatic solution.

Ideally, I would like to take a temporary transcription of the machine and edit it manually.

Answer 1

CMUSphinx can do this, just run the Aligner demo from sources. The latest sphinx tutorial is available in the sphinx4 tutorial . In the summer of 2014, CMUSphinx integrated continuous sound equalization into the sphinx4 trunk.

This is completely wrong, but unlike the voice base, it captures at least a few words. It is probably easier to decrypt it manually and set a time later. CMUSphinx can also do this by aligning audio to text.

There are quite a few problems with decoding accuracy caused by accents and reduced time, but there are things for which you are also responsible. For example, you should never convert 8khz bandwith phone sound to mp3. This conversion significantly reduces the accuracy of automatic speech recognition and overall recording quality. You can use other lossless speech codecs for compression.

Other questions discuss the topic, listing some recommendations.

Answer 2

When you search on Google, you will get a long list of transcription programs that works from market professionals. Some examples:

• Express scribe
• Eurescribe
• Inqscribe

But the problem arose when using automated software that it does not give the exact results that we need. Therefore, if you are still having problems using this software, you can hire a transcript to facilitate your work.

Answer 3

For this purpose, a new software (chrome application) was created called Speechlogger ( https://speechlogger.appspot.com ). It saves time stamps and allows you to export to srt format (captions) .

This is supposed to be the best technology because it uses Google’s speech recognition engine. It also automatically emphasizes (based on the simple content and structure of words).