Looking for a quick and easy way to transcribe your audio or video files? You've come to the right place!
In this article we suggest you the best practices to obtain a quality transcription using speech recognition technology. First of all, it should be pointed out that currently there is no automatic transcription technology able to transform all types of audio into text, especially those of poor quality, that we will see below. Fortunately, there are tricks to be followed to get a transcription 100% faithful to the original audio.
How to make a good quality audio recording?
And if you already have an audio recording, which Authôt services should you choose to have a perfect transcription?
For that, a few steps to follow: from audio recording to transcription, join the Authôt adventure!
Acoustics: Try to move away from the walls when making a recording. It causes what is called the "rebound effect", which generates a reverberation and gives the impression that your voice is far away. It is best to look for a place with sofas and curtains, as they keep out outside noise and also create the "studio effect".
Speech: It is important to articulate well when you are recording. However, do not speak in a jerky way, just have a regular and smooth speech tone, without rushing. You can be several speakers during the recording, each of them must speak in turn and not to cut themselves off.
Digital recorder: Record your audio file with a digital recorder. At Authôt we work in partnership with the professional brand NAGRA. We invite you to discover our bundled offer here. Although smartphones offer quality dictaphone applications, digital recorders are highly recommended. Indeed, microphones that equip most digital recorders give excellent sound reproduction. Thanks to their arrangement, they pick up sounds coming from the front as well as those coming from the side. With a smartphone, you have to face the audio source and be very close to get it. In addition, the memory of a smartphone is smaller, and you need to activate the airplane mode to not receive sound notifications.
Microphone: Even if you are recording alone, it is always recommended to use a tie microphone. They offer excellent acoustic qualities. Recordings can be made during meetings with several speakers, in this case, it is recommended to have at least two microphones placed at the ends of the table and connected to a recorder in order to capture the words of each. It can also be tie microphone for each speakers. The room can also be equipped with a conference sound system. During meetings, people should not cut oneself off, or speak at the same time. This greatly alters the transcription.
Record all at once: Try not to stop recording each time you pause. It is necessary to keep a natural speech, the retouching can always be possible thereafter.
Edit: It is recommended to modify the record, and clean errors or silences. This with the objective of leaving a clear and pleasant sound to listen to. We suggest you to use Audacity, one of the best free software of its kind for editing audio recordings.
Example of good quality audio recording for automatic transcription
Quality accoustics: studio recording
Professional speech: Voice-over
High quality recording equipment: studio recording
If your audio recording have a good quality, Authôt's automatic transcription service is the perfect solution. You will get a transcript very quickly: 12 min for a 1 hour audio recording! Then you can easily modify errors made by the system (spelling, specific vocabulary) on the online editor using the karaoke function. A wide choice of export formats is available: .srt,.html,.docx,.txt,.timecode etc.
When you recorded your audio or video you weren't aware of advise set out above and your file is no good quality? No worry.
As a reminder, poor quality recordings for automatic transcription systems are audio/video files recorded with a non-professional device whose average definition lets unwanted noise through such as reverberation or surrounding background sound (e.g. mobile phone). These are also recordings whose speech flow is not smooth (ex: hesitant speech, accents, speakers who cut themselves off). This can also mean that the speaker may be far from the mic, have a very strong accent or a bad speech. It can also be the case of a conference where several people speak at the same time.
The transcription is not infallible, so this kind of audio recordings complicate the system's work.
Example of poor quality audio recording for automatic transcription:
Is Kylian Mbappe the futur Pele?
Poor audio signal quality: noisy conversation, echo problems.
Disturbances in speech flow: strong accent, hesitant speech, superimposed speech (multiple speakers)
Audio with multiple languages: English and French
Although it's always better to have good quality sound as seen before, we know you can't always have it. That's why we offer a 100% human proofreading service! It allows us to accept this type of audio, which free online software cannot do, or more generally, automatic software or tools based solely on transcription technology.
The Authôt additional service is divided into two parts: correction and validation. Our proofreaders are all carefully selected: professionals and native speakers of the chosen language. Our proofreading service is handled internally by our team to ensure the delivery of quality work. So, if you choose proofreading, you will get a perfect transcript.
Despite a 95% reliability on a good recording, the choice of the proofreading service remains for many, a must have. Thus, to compensate for a lack of time or staff, users make the choice of human proofreading. This additional service allows you to obtain 100% reliable transcription without having to correct or proofread yourself the transcription done by the automatic system! This way, Authot ensures a follow-up of your projects from A to Z.
Now you're ready to turn your audio into text!
Authôt. You speak. We write.