FAQ – computer

We currently support:

To log into the online MOV recorder, use the same login/password than the one you use to login onto the ‘my-own-voice’ website.

You need to contact us via the ‘Start now’ section to get login credentials, and then choose your language and start the recording.

Please have a look to the custom message guidelines and tips

The MOV Recorder must have total control over the microphone function. On the latest Windows and MAC OS X operating systems, the default settings might possibly prevent the MOV Recorder from working properly.

On Windows 10 and higher, you may verify the following:

• In the settings panel (Windows + i) > Privacy > Microphone >
• Ensure that “Allow apps to access your microphone” and “Allow desktop apps to access your microphone” are both on the “ON” position

On MacOS 10.14 and higher:

• On your Mac, go to the Apple menu > System Preferences, click Security & Privacy, then click on Privacy.
• Select Microphone.
• Select the tickbox next to the “Acapela Recorder” app to allow it to access the microphone.

The better the recording quality – the better the resulting computer voice.

An ideal set up would be a professional studio, and someone monitoring your pronunciation.

However, if you do not have the possibility to record in a professional studio, you can do the recording in a quiet room at home at your convenience.

Avoid street and traffic noise, and remove any other source of background noise (e.g. ticking clocks, mobiles, barking dogs, noisy neighbors etc.) to avoid getting them into your recordings.

From the Options tab of the Acapela recorder, you can:

To complete the recordings, you will require a good Internet connection and a computer.
You will need a good quality directional microphone (see suggested microphone specifications below). The advantage of a head-mounted microphone is that it allows you to keep more or less the same distance.

Just keep in mind not to breath directly into the microphone.

We do not advise recording with a built-in laptop microphone, because:

• It is difficult to keep a consistent distance from your laptop microphone and to maintain good posture as you speak.
• Built-in microphones are usually omnidirectional and designed to pick up sounds from everywhere, thus also ambient noise!
• They are often of inferior quality compared to the head-mounted ones.

As explained in the “What microphone should I use?” section, we always suggest using a head-mounted microphone (headset) to keep the distance between the microphone and mouth constant.
Regarding quality, we suggest that the microphone meets the following specifications:

The recordings can be done from home through our website interface with a headphone set. You can try using a quality directional headset microphone (like e.g. a Sennheiser USB headset, Logitech USB Headset).

Stick to the same distance from your mouth to the microphone and maintain computer volume settings as in the previous sessions.

Avoid making breathing sound directly into the microphone to prevent clipping.

A Bluetooth microphone doesn’t have the sound quality that a corded microphone does and it is not recommended for the usage of the service.

Although it may be acceptable for instances when cords are impractical, such as when recording on the move or when you are at a long distance from your recording device. This often results in disconnections and therefore, alters the quality of the recordings made.

You will need reading materials that reach a volume of 50 sentences to create a voice. The sentences should be phonetically balanced and contain all the sound combinations necessary to synthesize the target language. The synthetic voice will be capable of reproducing any written text.

Yes, you can upload a txt file before starting the recordings with up to 300 personal phrases or expressions to be recorded. Those phrases will be recorded after the Acapela 50-phrase script is recorded. Note that the audio files will be delivered with your synthetic voices package and not as separate audio files.

Those messages (also called banked Messages) are personal and private audio messages created for you and about you, via the my-own-voice service. These messages make it possible for loved ones to keep a connection with you using more emotions and expressions together with your synthetic voice.

IMPORTANT NOTE: The custom messages need to be set and uploaded to the my-own-voice website before you start the voice banking process for the best experience.

More information about the process in the Upload of custom messages material procedure manual.

The number of sentences you will need to record is 50. It may take between 20 to 30 minutes to record the complete set, depending on how efficient you are.
The total set of sentences doesn’t need to be recorded in one session, but can be split over different sessions depending on your available time and voice constraints.

You can record the 50 phrases (or more if you add custom messages) in one session (average between 20-30 minutes) or with multiple sessions, but within a short time interval (e.g : 2 sessions of 15 minutes within 2 or 3 days).
It is always recommended to make the recording within a short time interval to avoid changes in the voice. Our statistics show that recordings completed within 2 or 3 days provide optimal results.

The best scenario is to record all in one session.

Here is a list of tips for the recording session and on how to improve the voice quality:

• Try to make volume/clipping controlled recordings and use a good quality microphone.
• Keep to the pronunciation in the sound example for each sentence and keep a constant speech rate and voice volume throughout the sessions.
• Well-articulated and steady speech and breathing may give better segmentation results, but keep the pronunciation natural by all means (do not over pronounce separate sounds or syllables).
• Keep the same pace and a similar pitch level, intonation, speech rate and voice quality as in previous sessions. E.g. If your nose is blocked, you will sound more nasal. Listen to your own recordings from previous sessions prior to the start of the new one and stick to your own voice and microphone settings from the previous sessions.
• Be aware that your TTS will sound relatively similar to the way you choose to speak.
• Exaggerated or very “jumpy” intonation might produce a somewhat mediocre synthesis result.
• You should read a few sentences before the real recording starts, to warm up the voice.
• Avoid big meals during the recordings. Avoid strong sugary or salty snacks/drinks so as not to dry out your vocal chords or cause an abnormal production of mucous that might impair your way of speaking.
• Take enough breaks, drink room-temperature water (before you are actually thirsty), and do not push your voice to the limits (if you are untrained, the voice quality might change as you get tired).
• Warm up your voice, and keep going, while you are “in the flow”.
• Clearing your voice can help, but if you find yourself doing this too often, take a small break or drink some water.

There are several steps/tips to consider: Keep the same mouth-to-mike distance and computer volume settings across different sessions.

• Calibrate your microphone at the start of the Acapela recorder (A. in the picture below)
• A green light indicates the sentence has been recorded, a yellow light indicates the ongoing recordings and the red light indicates that the sentence has to be recorded. (B. in the picture below)
• You can listen to what you have recorded by pressing C. (available if the file has been saved)
• Listen to the example of standard pronunciation in D.

Examples of the recording interface:

Since October 2020, there is no audio quality check anymore from Acapela.
We encourage the user to ensure the audio quality is sufficient, while the first three recordings are played back out loud (for each new recording session initiated).

Here we state the most important aspects to consider for a successful recording. More details and tips are provided in the rest of the FAQ.

• Use a headset microphone to keep the mouth-to-microphone distance constant throughout the recording, and use the same microphone for the entire recording.
• If possible, make all the recordings in the same room in order to maintain a constant acoustic and background noise during the whole recording.
• Always use the calibrate function before each session.
• Speak normally in a relaxed way, do not over-articulate or use exaggerated intonations.
• After pausing the recording, listen to some of the previous recordings before starting with new ones, to make sure you continue with a similar tone and intonation.
• Take pauses whenever you feel tired, and drink water regularly during the recordings.

Listen to the pronunciation example on the recording interface by clicking on the sentence (D):

The better your pronunciation matches the pronunciation “expected by the system, the better a TTS (Text-to-Speech) version (synthetic voice) you will get. “Stick to the punctuation in the sentence: i.e. make pauses, where there is punctuation (e.g , ; :), and avoid pauses in other places. recorded as it is.

Q: What if my accent is different from the sound example in the sentence? A: Normally a different accent even a non-native accent should not be a problem for a “correct” reproduction of your voice as it is, as long as you are more or less consistent in the difference to the main accent. Some differences, affecting the order of sounds may turn out more disruptive than inherent sound characteristics : such as e.g.

• pronouncing intrusive sounds in English “law-r-an-order”.
• or choosing a different variant pronunciation than the one assumed by the system (e.g. often : [O1 f @ n ] vs. [O1 f t @ n ]).
• Such intrusive/reordered sounds might inadvertently show up in the resulting synthesis in places where you do not expect them.

If you are not sure about how to pronounce a chain of sounds or a word in a sentence, listen to the pronunciation example for the sentence and copy it.

For a more realistic test we recommend testing the demo using the speaker and not the headphones, as this will be the way the product will be used. We also encourage testing using portable units like tablet, smartphones and other internet-connected devices to get as close as possible to the real experience when testing the voice via the demo.

The material uploaded by each voice created is stored and used only during the creation process. All materials are deleted from our servers once the voice is created and is made available for listening online. It is the user’s responsibility to ensure that the material uploaded is backed up and saved properly.

The voice created is stored for a period of one year in our system (legal commitment). On the practical side, this does not necessarily mean that we would delete the material, unless the patient or relatives instruct ACAPELA to do so.

However, partial recordings of an uncompleted script will only be stored for one year.

The FAQ on this page is mostly concerned with technical aspects of the recording, for a complete production description please visit the  learn more page of the my-own-voice website.