Collect, validate, and benchmark speech data for African languages to build better voice AI. Public TG https://xitter.cf/t.co/IkVeY03wTN
Africa
Joined October 2022
- Tweets1.2K
- Following7
- Followers4.3K
- Likes2.4K
Dialectra retweeted
If you're working on ASR, TTS, or conversational AI for African languages, come find us on Hugging Face. The data is there.
Let's build together.
Today, Dialectra releases its first open Hausa voice dataset on Hugging Face for free.
This is a small but meaningful step toward something bigger: dialect-aware AI infrastructure for African languages.
Hausa is one of the most widely spoken languages in Africa, with over 70 million speakers across Nigeria, Niger, Ghana, and beyond. Yet most voice AI systems fail Hausa speakers not because the technology can't work, but because the right data doesn't exist.
Dialectra is building that data layer.
This first dataset includes:
→ Dialect-tagged audio recordings from native Hausa speakers
→ Rich metadata: region, speaker profile, speaking style, quality score
→ Benchmark-ready structure for ASR and TTS development
We're making it available to researchers, AI teams, and developers who are building voice technology that actually works for African users.
This is just the beginning. More languages, more dialects, more infrastructure coming soon.
Dataset: huggingface.co/datasets/Dial…
Learn more: Dialectra.io
Dialectra retweeted
This is a solid step in the right direction.
African AI progress will depend heavily on datasets like this - not just models. Hausa being properly represented means real users can finally be included, not ignored.
If this momentum continues, we’re not just building AI for Africa… we’re building AI with Africa.
Today, Dialectra releases its first open Hausa voice dataset on Hugging Face for free.
This is a small but meaningful step toward something bigger: dialect-aware AI infrastructure for African languages.
Hausa is one of the most widely spoken languages in Africa, with over 70 million speakers across Nigeria, Niger, Ghana, and beyond. Yet most voice AI systems fail Hausa speakers not because the technology can't work, but because the right data doesn't exist.
Dialectra is building that data layer.
This first dataset includes:
→ Dialect-tagged audio recordings from native Hausa speakers
→ Rich metadata: region, speaker profile, speaking style, quality score
→ Benchmark-ready structure for ASR and TTS development
We're making it available to researchers, AI teams, and developers who are building voice technology that actually works for African users.
This is just the beginning. More languages, more dialects, more infrastructure coming soon.
Dataset: huggingface.co/datasets/Dial…
Learn more: Dialectra.io
Dialectra retweeted
Most people talk about model accuracy when it comes to AI.
We spend a lot of time thinking about what happens before the model ever sees the data.
At Dialectra.io our WER pipeline isn't just a final evaluation score.
Every voice submission goes through the following..
→ transcription
→ human review
→ normalization
→ annotation
→ verification
→ benchmark testing
This helps us catch errors early and continuously improve dataset quality before training begins.
The goal isn't simply to achieve a lower WER.
The goal is to understand "WHY" errors happen across different dialects, accents, speech patterns, and conversational contexts.
The infrastructure layer matters more than most people think.
Dialectra retweeted
Founders supporting Founders.
Drop what you are building, I will help you in users, feedbacks, posting a breakdown about the project/product, and of course funds
I will go first Dialectra.io
👇👇
Dialectra retweeted
Social authentication is live. You can now signup and sign-in using your gmail.
Next X and other Social handles.
Try it here. Dialectra.io
Dialectra retweeted
Dialectra (@_dialectra) has officially launched its Yoruba speech data campaign.
Following strong community participation across Hausa, Kanuri, and Fulfulde, Yoruba contributors can now record speech, transcribe audio, and join live conversations through Dialect Connect.
If you speak Yoruba, join the movement. Contribute your voice, earn rewards, and help build the future of African speech AI.
Get started today.
Dialectra retweeted
Submitted a proof of personhood for an accelerator while in a hospital bed… all for cloud credits to keep building as a bootstrap founder, we are really build from everywhere 😄
It's a really interesting journey about @_dialectra
Dialectra retweeted
Today I'm going to talk about @_dialectra
Not in English, but in Hausa.
As someone who uses AI almost every day and spends a lot of time exploring AI tools and projects, I noticed something interesting.
Yawancin AI voice agents suna iya magana da Hausa, amma idan ka zurfafa ka duba, da yawa daga cikinsu ba sa fahimtar yadda Hausawa ke magana a zahiri.
Hausar Kano daban.
Hausar Katsina daban.
Hausar Sokoto daban.
Har ma kalmomi, karin magana da lafazi suna canzawa daga yanki zuwa yanki.
Anan ne Dialectra ta bambanta.
Maimakon su mayar da hankali kawai wajen gina AI mai magana, suna tattara sahihin bayanan murya daga masu magana da Hausa na gaskiya.
Ba wai karatun rubutu kawai ba.
Suna tattara yadda mutane ke magana a rayuwa ta yau da kullum, da lafazi, da karin magana, da bambancin yare daga yankuna daban-daban.
Wannan yana da muhimmanci saboda AI ba zai iya fahimtar abin da bai taba koya ba.
Idan bayanan da aka horar da shi da su ba su wakilci Hausawa na gaskiya ba, to ko da model ɗin ya yi ƙarfi, zai yi kuskure idan ya gamu da ainihin masu amfani.
Abin da ya fi daukar hankalina shi ne cewa Dialectra tana gina foundation ne, ba kawai wani voice AI app ba.
Yau muna magana da @ElevenLabs, @Hailuo_AI da sauran voice AI platforms.
Amma ka taba tunanin me zai faru idan irin waɗannan manyan platforms suka samu damar amfani da ingantattun bayanan Hausa da Dialectra ke tattarawa?
Me zai faru idan AI zai iya gane Hausar Kano, Katsina ko Sokoto ba tare da rikicewa ba?
Me zai faru idan AI zai iya fahimtar yadda Hausawa ke magana a zahiri, ba kamar yadda littafi ya rubuta Hausa ba?
A ganina, wannan shi ne babban abin da ya sa Dialectra ta bambanta.
Ba wai kawai tana gina AI ba.
Tana gina bayanan da za su taimaka wa AI fahimtar Hausa yadda ya kamata.
Kuma hakan na iya zama babban mataki ga Hausa da sauran harsunan Afirka a duniyar AI.
Dialectra retweeted
If you want to learn more about Dialectra and why US..
Don't miss out this AMA session.
Dialectra retweeted
We launched "Dialect Connect" yesterday and in just 24hrs, the stats is really impressive.
I was thinking few days ago a simple idea: what if we could capture how people actually speak, not just how they read?
I then implements yesterday as a additional feature for Dialectra.io
24 hours later:
📞 371 conversation requests
✅ 303 completed conversations
🎙️ 45.1 hours of conversational speech collected
⏳ 5 pending
🟢 2 active
❌ 61 rejected
For years, most speech datasets have been built around scripted recordings. They are useful, but they only tell part of the story.
Language lives in conversations.
It lives in pauses, interruptions, storytelling, laughter, code-switching, local expressions, and the unique rhythm that makes every dialect different.
The future of voice AI will not be built solely on people reading sentences from a screen. It will be built on authentic human interactions.
That is what excites me most about these numbers.
In just 24 hours, hundreds of people chose to connect with complete strangers or friends and simply talk. In doing so, they generated something incredibly valuable: real-world conversational data for African languages and dialects.
Every completed conversation moves us closer to a future where AI can understand not only what Africans say, but how we say it.
When we started Dialectra, our mission wasn't just to collect voice data. It was to ensure that African languages, dialects, and identities are represented in the AI systems that will power the next generation of technology.
45.1 hours is a small number compared to where we're going.
But it's a reminder that the infrastructure for African voice AI won't be built in a lab alone. It will be built by communities, contributors, and everyday conversations happening across the continent.
We're still very early.
Dialectra retweeted
Africa is home to some of the world’s most spoken and culturally influential languages, yet modern AI systems still struggle to understand them accurately.
Hausa alone is spoken by an estimated 80 to 100 million people across West and Central Africa, particularly in Nigeria and Niger. Swahili, widely recognized as Africa’s leading lingua franca, connects more than 200 million speakers across East and Central Africa.
Arabic, one of the continent’s most dominant languages, is spoken by hundreds of millions across North Africa and parts of the Sahel, shaping commerce, education, religion, and communication throughout the region. Yet despite this enormous linguistic scale, African speech remains heavily underrepresented in global AI systems.
That is the gap @_dialectra is stepping in to solve building the speech infrastructure designed to help artificial intelligence truly understand how Africa speaks.
Dialectra retweeted
We’re currently preparing to launch our first major dialect campaign together with our partners at Dialectra.
This next phase will focus heavily on:
• Hausa
• Fulfulde
• Kanuri
If you speak any of these languages or dialects, now is the best time to sign up and get ready for what’s coming.
We’re building more than datasets we’re building the foundation for dialect-aware AI systems across Africa.
A lot more updates before the end of the month..
Dialectra retweeted
Being a #Muslim founder is more common than we know, it just hasn't been mapped yet. This is why we are proud to announce the 1st Muslim #Founder Directory. Over 200+ Muslim founders 💪 Its time we start competing on every vertical.
Check it out ⭐️ ummah.build/founders
Dialectra retweeted
All submissions, all verification, all recordings are counted and moat be rewarded at the end..
Still day one
Dialectra retweeted
Indeed @elonmusk you have to look into what @_dialectra is built. It will indeed help solve this problem.
This Post is from a suspended account. Learn more
Dialectra retweeted
Everyone is talking about Dialectra…
But most people still don’t understand what it actually is.
Here’s a simple breakdown 🧵👇
Dialectra retweeted
We recently launched our transcription feature, and we’re already seeing strong momentum with 18K+ voice datasets submitted so far.
@_dialectra now has two active data collection pipelines:
• corpus script reading
• transcription
This allows us to capture both structured speech and real-world language usage a critical step for building better dialect-aware AI.
We’re still at day one....
If you’re building models for dialect-rich languages especially Hausa, Fulfulde, or Kanuri, we’re open to collaborate and support with high-quality data.
Let’s build.
Dialectra retweeted
You don’t need to speak a language to train a good AI model.
What matters is the source of your data.
Most speech models fail not because of the engineers building them but because the datasets don’t capture how people actually speak.
At Dialectra.io we focus on that layer, If you train on our datasets, even as a non-native or non-Nigerian, your model can learn:
• native Hausa tone
• accurate pronunciation
• natural rhythm and speech patterns
Because the intelligence is in the data not just the model.
This is why dialect-aware, high-quality datasets matter.
Build with the right data, and the model follows.