3650HQ automatically transcribes every WhatsApp voice note in your inbox to text, in English, Yoruba, Hausa, Igbo, Swahili or Zulu. Skim a customer's order details in seconds instead of replaying three minutes of audio in a noisy shop.
Transcription covers English, Yoruba, Hausa, Igbo, Swahili and Zulu; Nigerian Pidgin isn't supported yet. It transcribes what was said; it doesn't translate between languages or summarize long threads. It costs 0.05 credits per voice note, the same as replying to the customer who sent it, and a fraction of the 2 credits it costs to start a new conversation with a template or campaign.
Typing is slow, keyboards are small, and a customer describing what they want in their own language does it in ten seconds of audio rather than three minutes of thumbs. In a lot of Nigerian, Kenyan and Ghanaian DM commerce, voice notes are not a quirk, they are the primary way customers place orders.
For the seller this is a worse deal than it looks. Audio cannot be skimmed. A voice note has to be played from the start, at the pace the customer spoke, with the phone against your ear, which is impractical in a market, a shop, or anywhere else with noise. You cannot search it. You cannot glance at it to check an address before dispatch. And when a customer sends five of them in a row, the only way to find the one with the delivery address in it is to replay all five.
None of that gets better with volume. The busier the business, the more expensive listening becomes.
3650HQ transcribes voice notes automatically as they arrive. The transcript drops in directly beneath the audio player in the same conversation, so the audio is still there if you want it and the text is there if you do not.
An order, a size, a colour and an address can be read at a glance. Checking one detail before dispatch stops being a ninety-second task.
Once a voice note is text, it is part of the thread's content rather than an opaque attachment sitting in it.
A voice note that stays audio is invisible to everything downstream. Transcribed, it becomes context an AI draft can work from, so a reply suggested on a voice-note conversation is based on what the customer actually said.
The customer records a voice note the way they always have and does not know anything happened. There is no app, no setting for them to enable, and nothing about their experience changes.
Transcription covers English, Yoruba, Hausa, Igbo, Swahili and Zulu. That set is the actual point of the feature: general-purpose WhatsApp inbox tools built for other markets transcribe English and stop, which is not much help when the customer placing the order is speaking Yoruba.
Nigerian Pidgin is not supported yet, and it is worth saying that plainly rather than hoping nobody tests it. Pidgin is extremely common in exactly the conversations this feature is meant to serve, so its absence is a real limitation and not a rounding error.
3650HQ also does not translate. A Yoruba voice note becomes Yoruba text, not English text. Transcription and translation are different problems, and quietly turning one into the other would mean a seller reading a paraphrase of what their customer said without knowing it.
Transcription costs 0.05 credits per voice note, which is the same as sending one reply to the customer who sent it, and a small fraction of the 2 credits it costs to open a new conversation with someone.
Because 3650HQ runs on prepaid credits rather than a subscription, this scales with how much your customers actually talk. A week where nobody sends a voice note costs nothing, which is a different shape of bill from a monthly per-agent fee that arrives whether or not anyone used the feature.
Yes. 3650HQ automatically transcribes inbound WhatsApp voice notes to text in English, Yoruba, Hausa, Igbo, Swahili and Zulu, right inside the conversation.
Yes. Transcription runs on voice notes received through your connected WhatsApp Business (Cloud API) number inside 3650HQ's inbox.
Not yet. Transcription currently covers English, Yoruba, Hausa, Igbo, Swahili and Zulu.
0.05 credits per voice note, a small fraction of a credit, and free credits are included when you sign up.
No. Transcription happens automatically for every voice note that arrives in your connected WhatsApp inbox, nothing to configure per message.
Directly under the audio player in the same conversation. The original voice note stays where it was, so you can still play it if the transcript is ambiguous.
No. A Yoruba voice note becomes Yoruba text. Transcription and translation are different things, and turning one into the other silently would mean you were reading a paraphrase without knowing it.
Nothing changes on their side. They record a voice note as usual; there is no app, prompt or setting involved for them.
Yes, once it is transcribed. Audio on its own is invisible to everything downstream, so transcription is what lets an AI-drafted reply be based on what the customer actually said.